ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2018

Digital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Medicine 1473

Steering health and social carethrough quasi-markets

DAVID ISAKSSON

ISSN 1651-6206ISBN 978-91-513-0368-0urn:nbn:se:uu:diva-354476

Dissertation presented at Uppsala University to be publicly examined in B21, BMC,Husargatan 3, Uppsala, Friday, 7 September 2018 at 13:15 for the degree of Doctor ofPhilosophy (Faculty of Medicine). The examination will be conducted in Swedish. Facultyexaminer: Professor Anders Anell (Lund University, Department of Business Administration).

AbstractIsaksson, D. 2018. Steering health and social care through quasi-markets. DigitalComprehensive Summaries of Uppsala Dissertations from the Faculty of Medicine 1473.52 pp. Uppsala: Acta Universitatis Upsaliensis. ISBN 978-91-513-0368-0.

Municipalities and county councils try a multitude of different strategies when they design andsteer health and social care markets to ensure that goals such as quality and equity are met.Depending on the strategies used, different problems arise. The aim of this thesis is to examinehow local authorities can design quasi-markets in a way that achieves public goals such as equityand high quality. To answer the aim, four empirical studies were carried out.

The studies show that when designing a market by contracting-out through publicprocurement, the issues lay primarily at specifying and defining what is meant by quality beforea service is privatized. This is especially difficult to do concerning soft areas such as elder-and healthcare. If this is not done properly, it can lead to crucial issues for monitoring qualitysince the contracting authority cannot hold the provider responsible for delivering an aspect ofa service if that aspect is not specified in the contract.

When a market is designed as in the patient choice systems in primary care, it creates awhole other set of difficulties for the local governments. Here, it is not as important to specifyquality beforehand in the contracts since quality monitoring is done retrospectively by boththe counties themselves as well as the patients who with their choices can monitor quality bypunishing providers with poor quality by registering with another provider. Instead, the crucialproblem is how to design reimbursement system that will lead to an equal access to health care.In this respect, the county councils utilize different methods. However, despite these measures,the primary care choice reform have led to inequity, both geographical inequity in regards towhere new private primary health care centres are located but also, to a larger degree, socio-economic inequity relating to what kind of socio-economic groups of individuals are registeredwith private PHCCs. In other words, county councils do not manage to fully counteract riskselection behaviour by the design of their reimbursement system which could imply issues withunequal access to health care.

Keywords: patient choice, contracting, equity, health policy, public procurement, quasi-markets

David Isaksson, Department of Public Health and Caring Sciences, Health Services Research,Uppsala Science Park, Uppsala University, SE-75185 Uppsala, Sweden.

© David Isaksson 2018

ISSN 1651-6206ISBN 978-91-513-0368-0urn:nbn:se:uu:diva-354476 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-354476)

To Martina, Tage and Stellan

List of Papers

This thesis is based on the following papers, which are referred to in the text by their Roman numerals.

I Isaksson, D., Blomqvist, P., & Winblad, U. (2017): Privatization

of social care delivery – how can contracts be specified?, Public Management Review, DOI: 10.1080/14719037.2017.1417465 Published online 27th December 2017.

II Isaksson, D., Blomqvist, P., & Winblad, U. (2016). Free estab-

lishment of primary health care providers: effects on geograph-ical equity. BMC Health Services Research, 16(28), 1–15. http://doi.org/10.1186/s12913-016-1259-z

III Isaksson, D., Blomqvist, P., Pingel, R. & Winblad, U. Risk se-

lection in primary care – a cross-sectional fixed effect analysis of Swedish individual data. (Submitted)

IV Winblad, U., Isaksson, D., Blomqvist., P. Preserving social eq-

uity in privatized primary care: a matter of contract design. (Manuscript)

Reprints were made with permission from the respective publishers.

Contents

Introduction ................................................................................................... 11 

Aim and research questions .......................................................................... 14 

The Swedish social- and health care systems ............................................... 15 National goals of the health and social care systems ............................... 15 Development of privatized social and health care in Sweden .................. 16 

Public procurement and competitive tendering ................................... 17 Vouchers and user/patient choice systems ........................................... 18 

How can local governments achieve their goals through contracting? ......... 20 Contracting and incomplete contracts ...................................................... 20 Choice and equity ..................................................................................... 22 

Methods and findings .................................................................................... 25 Study I – Contracting out care homes ...................................................... 25 

Results ................................................................................................. 26 Study II – How does patient choice affect geographical equity ............... 27 

Results ................................................................................................. 28 Study III – Risk selection in Swedish primary care ................................. 28 

Results ................................................................................................. 29 Study IV – Risk adjustment measures in Swedish primary care .............. 29 

Results ................................................................................................. 29 

Discussion ..................................................................................................... 31 Principal findings ..................................................................................... 31 Methodological considerations................................................................. 33 

Theoretical and policy implications .............................................................. 36 

Further studies ............................................................................................... 39 

Summary ....................................................................................................... 40 

Summary in Swedish .................................................................................... 41 

Acknowledgements ....................................................................................... 44 

References ..................................................................................................... 46 

Abbreviations

ACG Adjusted Clinical Groups CNI Care Need Index NPM New Public Management PCCR Primary Care Choice Reform PHCC Primary Health Care Centre SES Socio-economic status

11

Introduction

Two of the main principles guiding the Swedish health care system are soli-darity and equity, with health care services primarily financed publicly via taxes, and resources allocated according to medical need (Anell, Glenngård, & Merkur, 2012; von Otter & Saltman, 1991). Beginning in the 1980s, far-reaching reforms within the public sector were initiated in Sweden, as in many other countries. The reforms, which sought to make the public sector more cost-effective, were inspired by a set of ideas which have become known as New Public Management or NPM (Gruening, 2001; Christopher Hood, 1991; Pollitt & Dan, 2011). NPM originated in a criticism of the bureaucratic, cen-tralised government and a desire to reduce the ever-increasing costs of the public sector. The basic idea behind NPM is to use mechanisms similar to those found in the private market in order to increase efficiency by reducing costs and increasing quality (Christopher Hood, 1991; Walsh, Deakin, Smith, Spurgeon, & Thomas, 1997). One of the most notable ideas that sprung from the NPM movement is the privatization of welfare service delivery such as health and social care. The privatization of the production of health care can be achieved through several methods such as contracting or by introducing so-called patient choice sys-tems. Patient-choice systems are based on the principle that patients them-selves choose between competing care providers – both public and private – within a publicly regulated and financed system. Both contracting and patient choice systems have been introduced in Swedish health and social care. Since the early 1990s, Sweden has seen an ongoing movement towards marketization of the welfare system where private health and social care providers have been encouraged to enter and compete with public providers. The introduction of competition and private, including for-profit, providers in health and social care has been considered by some to be a threat to the Swedish health care system’s principles of need-based, equal access to health care services for all citizens. One concern has been that private providers will focus on the more profitable groups of patients, e.g. those with the lowest health care needs, and neglect other groups (Burstrom, 2010; Dahlgren, 2008).

12

Contracting and patient choice systems can both be viewed as forms of quasi-markets, i.e. systems with features of both a hierarchical bureaucracy and a conventional market (Fotaki, 2007; Le Grand & Bartlett, 1993). Quasi-mar-kets are generally understood as a mix between hierarchical steering and tra-ditional markets (Kastberg, 2008; Le Grand & Bartlett, 1993). The creation of quasi-markets pose new challenges for policy makers with respect to govern-ance (Boyne, 1998; Donahue, 1989). When governing health care systems, politicians and bureaucrats need to ensure that laws are followed and public goals such as equity and high quality are met. In a fully public system such as that in Sweden during the 1970s, organized as a Weberian bureaucracy, poli-ticians and civil servants can micromanage the production and distribution of medical services. When a public organisation privatises – wholly or in part – the production of care services, the system increasingly relies on the market forces and contracts to achieve desired outcomes. To achieve public goals such as high quality or equal access to health care for all citizens, the public authorities responsible for health and social care (e.g. municipalities or county councils), must create the right incentives for health care providers (Le Grand & Bartlett, 1993; Puig-Junoy, 1999; Sappington & Stiglitz, 1987). This can for example be done through specification of quality criteria that providers must follow, and through financial reimbursements that rewards certain be-haviour. The intricate task for the public authorities managing a quasi-market, however, is to construct these incentives in a way that makes it possible to monitor, or verify, the desired behaviour on part of the providers. When it comes to the goal of quality, the first challenge is to define what is meant by the concept. The second challenge is to formulate quality criteria that both cover all aspects of the health service, but are also possible for a public au-thority to monitor. If quality criteria are not able to be monitored, economic incentives are created for the providers, especially profit-driven ones, to de-liver services with an inferior quality, i.e. the quality-shading hypotheses (Domberger & Jensen, 1997). In the case of equity goals, the challenge for public authorities trying to steer a quasi-market in health care, is to ensure that services are delivered according to need. Another way of steering quasi-markets is through different forms of reim-bursement systems. In order to ensure high quality care, the public authority must design the reimbursement system in such a way that it creates incentives for the providers to provide high quality care. The greatest challenge here is that regardless of how a local government designs a system, health care pro-viders have the possibility to adapt their behaviour in response to that reim-bursement, thus maximizing their profits (Koning & Heinrich, 2010). In a more general sense this issue has been well-known for a long time and its essence is captured in Goodhart’s law: “When a measure becomes a target, it ceases to be a good measure” (Elton, 2004). Within health care, an important

13

goal is equity in the delivery of services. One of the main concerns with re-gards to marketization and equity in health care is the behaviour that arises when providers adapt to economic incentives by prioritizing some patient groups and excluding others. This have been referred to in the literature as risk selection, or creaming (Ellis, 1998). Creaming here refers to an over-provision of treatment to patient’s that are possible to treat at low-cost (Ellis, 1998).

Irrespective of what type of quasi-market is being utilized, this organizational form implies that a new way of governing is introduced for politicians and bureaucrats. The challenge for governments engaging in this form of govern-ance lies foremost in carefully deciding when to marketize, or privatise, a ser-vice, and when that it is done, what mechanism to use to achieve desirable political goals. How this is done in practise by public authorities and what effect it has on goals such as quality and equity is the topic of this dissertation. The dissertation examines these processes through the Swedish case, which in several ways is distinct from other national health care systems. The Swedish health care system is characterized by a high degree of decentralisation, with the primary responsibility for health care governance delegated to 21 local governments. These local governments are directly elected, and have the re-sponsibility for the delivery and financing of health care.

14

Aim and research questions

Steering through markets creates new challenges for local authorities. By de-signing relevant and explicit quality criteria and reimbursement systems, they try to create market conditions that lead to desirable outcomes. The aim of this thesis is to examine how local authorities have designed quasi markets to achieve public goals such as equity and quality. The main research questions in the thesis are:

How do municipalities and county councils design and steer health and social care markets to ensure that public goals such as quality and equity are met?

How does the design of the quasi-markets affect equity within health care?

To answer these research questions, four empirical studies were carried out. In the first study we examined contracts in elder care and measured to what degree the quality requirements were monitorable. The second and third study examined in different ways if a quasi-market introduced in Swedish primary care challenged equity in access to health care. The fourth study examined how local governments try to design the patient choice system to avoid risk selection of patients that can lead to inequity in access to health care.

15

The Swedish social- and health care systems

The Swedish social and health care system is highly decentralized compared with other nations. Both health care and social care are considered public re-sponsibilities in Sweden and are financed through state grants and local taxes levied by the county councils and municipalities (Anell et al., 2012). The gov-ernment is organised into a multi-level system consisting of the national gov-ernment, county councils and municipalities. The county council have the main responsibilities for health care while the municipalities are principally responsible for providing social services, including elder care (Anell, 2005). However, it is the national government through the Ministry of Health and Social Affairs who determine the general policies regulating health and social care, including formulating the main goals of the system. The national gov-ernment also provide state grants to the county councils, covering 15-20 per-cent of their health care expenditures (Anell et al., 2012).

National goals of the health and social care systems Two of the main goals of the Swedish health care and social care systems are equity and quality. One of the key paragraphs in the Health and Medical Ser-vices Act (SFS 2017:30) states that “The goal of the health care is good health and care on equal conditions for all citizens” (§ 3, SFS 2017:30). The next paragraph states that “[the care] should be of good quality with a good hy-gienic standard […]” (§ 3, SFS 2017:30). For social care, similar key paragraphs exists in the Social Services Act (§ 1, SFS 2001:453):

“The society’s social service shall on the basis of democracy and solidarity

promote citizens’: - Economic and social safety - Equity in living conditions - Active participation in social life”

These goals could be formulated by any level of government, i.e. national, regional or local, but irrespective of the principal of a certain goal the one responsible for carrying out these goals in practice are the county councils and municipalities (Anell et al., 2012). Thus, each county council or municipality

16

has the responsibility of organizing health and social care in a way such that these goals are reached. This is the case irrespective of whether the providers of health and social care are the local government itself or a third party such as a private company.

Development of privatized social and health care in Sweden Until the beginning of the 1990s, most health and social care in Sweden was provided by publicly owned provider units. Coinciding with the economic cri-sis in the beginning of the 1990s, Sweden, along with most parts of the western world, was influenced by the set of ideas which came to be known as New Public Management or NPM (Harrison & Calltorp, 2000; C Hood, 1995; Stolt & Winblad, 2009). One major change that took place under NPM was the introduction of pur-chaser-provider splits, entailing a separation of the purchasing and providing function of a government (Almqvist, 2001; Bevan & van de Ven, 2010; Tynkkynen, Keskimäki, & Lehto, 2013). This enabled different providers (public or private) to compete between each other. The separation of these two functions later facilitated the privatization of health and social care; when there already existed a separate arm of the government responsible for provid-ing care, the step to contracting out at least part of this provision was less problematic (Stolt & Winblad, 2009). Up until 1992, only around 1-2 % of health and social services were privatised. From then on, a steady increase in the share of privatized services has occurred (Jordahl & Öhrvall, 2014).

When discussing privatisation, it is important to distinguish between different forms of privatisation (Donahue, 1989). A simple distinction is between pri-vatisation of production, financing and regulation, see Figure 1. Privatisation of regulation means that something that has previously been regulated by the government now is unregulated or less regulated. Privatisation of financing occurs when a service that has previously been financed via taxes now is fi-nanced privately, for instance via insurance or out-of-pocket payments. Fi-nally, privatisation of production happens when a service that was previously produced by a publically owned unit is produced by a privately owned unit. Examples of ways privatisation of production can come about include public procurement and the creation of patient choice- or user choice-systems.

17

Figure 1. Typology over different forms of privatization

It should be noted that privatisation of production does not necessitate the pri-vatisation of financing or regulation. When contracting out health- or social care through public procurements or patient choice-schemes in tax-based wel-fare states, it is more common than not that financing still mostly occurs through taxes and the services are still to a high degree regulated by a national or local government (Blomqvist, 2005).

Public procurement and competitive tendering The introduction of NPM reforms in Sweden coincided with the preparation to enter the European Community. As part of the preparation, an EC directive on Public Procurement was implemented in the Swedish Act on Public Pro-curement (1992:1528). Thus, the legal framework for contracting-out social services and health care was set out which up until then legally had been a grey area (Bianchi & Guidi, 2010; Stolt & Winblad, 2009). The Act on Public Procurement (2016:1145) sets out a distinct process all local authorities must follow whenever they want to contract-out a good or a service. The local au-thority must first write contract documents that specifies what is being con-tracted out, the quality requirements and the criteria for selecting the winning tender. After this is done, all interested organizations may send in tenders which are then evaluated based on the selection criteria previously decided upon (M. Bergman & Lundberg, 2013; Forsberg, 2004). In other words, all quality requirements must be decided upon before the winning tender starts providing the service and the competition between companies generally takes place ex ante, i.e. before providing the services (Domberger & Jensen, 1997). The Act on Public Procurement (2016:1145) states that the contract must be provided to either the tender with the lowest bid or the tender that is econom-ically most advantageous. In the first case, the bidder that offers to perform a service for the lowest price wins the procurement as long as it fulfils all gen-eral requirements (M. Bergman & Lundberg, 2013). In the second scenario

18

the contracting authority must set up selection criteria and a formula on how to calculate an index used for deciding the winning tender. In Sweden, as well as in Europe, it has been increasingly common for contracting authorities to use a combination of price and quality for supplier selection (M. Bergman & Lundberg, 2013). Since the contracting authorities must decide upon a win-ning tender before the services are being produced, quality must be evaluated in ways other than by measuring the quality of the actual services provided. In health and social care, this is often done by letting the bidders describe how they will achieve good quality and then giving a quality score to each tender based on their description. The description of how a tender should work with quality later on becomes part of the contract and the winning provider is con-tractually obligated to follow it (M. Bergman & Lundberg, 2013; Lunander & Andersson, 2004).

Vouchers and user/patient choice systems An alternative to contracting out via competitive tendering and public pro-curement are user choice systems. The basic principle behind a user choice system is that the contracting authority authorizes providers without any guar-antee of a user base. All users may choose any of the authorized providers and the providers are consequently reimbursed based on how many users have chosen that provider (Forssell & Norén, 2013; Lundvall, 2012). User choice systems were introduced in Sweden in 1985, when a municipality wanted to let their users of choose their provider of medical pedicures (Winblad, Andersson, & Isaksson, 2009). During the 1990s local authorities came to use similar systems in primary education as well as medical and social care (Socialdepartementet, 2008; Winblad et al., 2009). In 2008, the Act on System of Choice (2008:962) was introduced, creating a clearer legal ground for local authorities wanting to establish user choice systems. The central gov-ernment wanted to promote this form of privatization and offered grants to municipalities wanting to create user choice systems within social care (Socialstyrelsen, 2010). Only two years later, it was made mandatory for all county councils to establish patient choice systems in primary health care (Winblad, Isaksson, & Bergman, 2012). The mechanisms behind user choice systems, which can be viewed as a form of contracting in that they build on contracts between local public authorities and private providers, are in several aspects different from the mechanisms behind public procurement. In a user choice system, all providers who meet certain basic requirements are allowed to participate on the quasi-market. Thus, the competition between different providers does not take place in the contracting situation but rather later as providers compete over patients (Forssell & Norén, 2013). In other words, user choice systems are based on

19

ex-post competition where the patients/users evaluate quality. At the same time, Swedish laws prescribe that it is the legal principal of a service (hu-vudman), i.e. the municipality or county council, that is still mainly responsi-ble for the quality of care services. (Andersson, Janlöv, & Rehnberg, 2014). This creates a difficult situation concerning how politicians and civil servants can monitor the services in a user choice system.

20

How can local governments achieve their goals through contracting?

The county councils have the main responsibility for health care whilst the municipalities are responsible for social care in Sweden (Anell et al., 2012). However, since Sweden is not a federal state, it is the national government that has the overarching responsibility for all general policies, including poli-cies focusing on health and social care. Thus the national government, through the Ministry of Health and Social Affair, can set out certain goals for the health and social care system (Anell et al., 2012). Two of the main national goals for health and social care are equity and a high level of care quality (Anell et al., 2012; Fredriksson, Blomqvist, & Winblad, 2013; Stolt, Blomqvist, & Winblad, 2010). Despite the goals being set out by the national government, it is the responsi-bility of the local authorities, i.e. the county councils and municipalities, to organize and carry out the care in a way that these goals are reached. The local authorities have this responsibility regardless of whether they produce the ser-vices themselves or if they choose to contract it out to another provider (Anell et al., 2012). However, when contracting out to another provider, some of the means by which authorities can govern are no longer there. This could create new obstacles for authorities when trying to achieve public goals such as high quality and equity.

Contracting and incomplete contracts A situation where a government contracts with a private party to deliver goods or a services can be described as a principal-agent relationship where the gov-ernment can be viewed as the principal that delegates tasks to an agent (Amirkhanyan, Kim, & Lambright, 2011; Eisenhardt, 1989). In this relation-ship, an integral problem is asymmetrical information between the parties where the agent generally has more information than the principal regarding how the operations are being run (Williamson, 1975). Thus, the relationship is susceptible to opportunism by the agent which creates (among others) two potential problems: adverse selection and moral hazard (Amirkhanyan et al., 2011). Adverse selection refers to hidden information in the initiation of the

21

contracting relationship, i.e. the principal does not receive correct information regarding the agent’s ability to meet its contractual obligations (Eisenhardt, 1989). Moral hazard, on the other hand, is related to hidden action where an agent has the possibility to take advantage of asymmetrical information and not fulfil its contractual obligations, for instance by ‘skimming’ (Arrow, 1985; Eisenhardt, 1989; Ellis, 1998). In other words, adverse selection relates to problems with asymmetrical information ex ante, whereas moral hazard re-lates to problems with asymmetrical information ex post. Contracts can be seen as a document which aim to specify the responsibilities between a purchaser and a supplier (Trevor L. Brown, Potoski, & Van Slyke, 2007). One method for trying to reduce the risks of opportunistic behaviour by the agents is to meticulously describe the terms and quality expected of an agent when carrying out a contract (Trevor L Brown, Potoski, & Van Slyke, 2006; Domberger & Jensen, 1997). However, research shows that contracts are imperfect tools for regulating the relationship between a principal and an agent (Trevor L. Brown et al., 2007; Slyke, 2013). Contracts containing every possible aspect of a relationship, i.e. complete contracts, can in practice be considered utopic and can never be achieved in practice due to imperfect in-formation and uncertainty about the future (Artz & Brush, 2000; Oliver Hart & Moore, 1999). It is basically impossible, or at least very costly, for a con-tracting principal to specify every imaginable or unimaginable situation which could arise during the term of the contract (Trevor L. Brown et al., 2007; Hendrikse & Veerman, 2001). The observation that the complexity of the real world makes it too costly, if not impossible, to describe all possible aspects in a contracting relationship is the foundation for the theory of incomplete contracts (Hendrikse & Veerman, 2001; Sclar, 2000). The incompleteness in a contract means that not all possi-ble situations can be described ex ante when constructing a contract which eventually may give rise to opportunistic behaviour by the agent ex post (Oliver Hart & Moore, 1999; Hendrikse & Veerman, 2001; Segal, 1999). When some aspects of a service is non-contractable, i.e. not specified clearly in the contract, it is difficult for the principal to prove that the agent does not provide the desired quality. As a result of the non-contractability, the agent’s incentive to reduce costs increases, which ultimately may lead to a reduction in quality, i.e. the quality-shading hypotheses (Domberger & Jensen, 1997).

22

Choice and equity Equity in health care can be defined as “the absence of systematic disparities in health between groups with different levels of underlying social ad-vantage/disadvantage – that is wealth, power, or prestige.” (P. Braveman & Gruskin, 2003). The definition above is an example of socio-economic equity. Another more specific form of equity regards access to health care. Equity in access can be defined as a fit between the demand for health care services among the population and the supply of care by providers; i.e., health care services need to be affordable, contain the appropriate treatments in relation to medical needs and be physically situated so that patients can get to them (Neutens, 2015; Penchansky & Thomas, 1981). The last point relates to geo-graphical accessibility, which is a concept that can be understood as the dis-tance or travel-time between patients and providers. Equity in geographical accessibility, or geographical equity, can be defined as all individuals having similar or a minimum travel-time to the nearest health care provider (Gravelle & Sutton, 2001; Luo & Wang, 2003; Newhouse, Williams, Bennett, & William, 1982; Rice & Smith, 2001). ‘Choice’ has been held out as an important value in social services like health and social care foremost after 1980, when the concept was popularized by thinkers like Milton Friedman (Blomqvist, 2004; Friedman & Friedman, 1980). ‘Choice’ as an organizational form in publicly regulated and financed systems of social services typically refers to the idea that public and private providers compete for users or patients within a publically financed system. Furthermore, the allocation of resources is based on the choices of the users which in theory will lead to improved quality since users will choose providers with the highest quality, thus providing them with more resources (Appleby, Harrison, & Devlin, n.d.; Dixon & Le Grand, 2006; Fotaki, 2010). Several scholars argue that choice and equity are two incompatible virtues (Bevan, Helderman, & Wilsford, 2010; Mendis, 2009; Oliver & Evans, 2005). A choice system is built upon the premise that people should decide what ser-vice provider is the best option for them. This notion can seem trivial but many argue that less educated and in other ways vulnerable groups do not have the capacity to make well informed choices (Burstrom, 2010; Dixon & Le Grand, 2006; Fotaki et al., 2005). If vulnerable groups are not able to make well in-formed choices, the consequence of this could be that they will receive health or social care of worse quality (Barr, Fenton, & Blane, 2008; Fotaki, 2010). Empirical studies give clear support for the idea that vulnerable groups with low socio-economic status indeed make fewer and less well-informed choices (Cheng, 2004; Dixon, Robertson, & Bal, 2010; Harris, 2003; Ringard, 2011; Robertson & Burge, 2011; Winblad et al., 2012). There is however a lack of

23

studies that have examined whether this have an effect on the actual quality of care (Winblad et al., 2012). Unequal possibilities to make well-informed choices are not the only issue that is discussed with regards to choice and equity. Another common theme in the literature is that of cream skimming. Cream skimming refers to the per-ceived practice of providers choosing users instead of users choosing provid-ers (Le Grand, 2009). The idea is that providers in different ways will try to attract patients that are profitable (Koning & Heinrich, 2010). Within health care the concept of cream-skimming has been captured in the terms creaming, skimping and dumping (Ellis, 1998). Creaming refers to an over-provision of treatment to patient’s that are possible to treat at low-cost, skimping is the conduct of under-provision of treatment to high-cost patients and dumping when a provider refuses to accept high-cost patients (Ellis, 1998). To prevent cream-skimming from arising, local authorities try to design their reimburse-ment systems in different way to counter this behaviour (Barros, 2003). Even though there exists a large theoretical discussion regarding the difficul-ties for governments to steer social services and health care through quasi-markets and contracts, relatively few empirical studies have examined sys-tematically how contracts are actually written or what effects different con-tract designs have on outcomes with regards to quality and equity. Patient choice systems are mainly phenomena that occur in tax-funded Beveridge type health care systems, and the literature examining the effects of patient choice will largely be limited to studies from NHS and the nordic countries. Some studies of the English NHS have investigated effects of increased com-petition and patient choice on equity. In one study that looked at the effect of competition on equity found that competition did not undermine equity (Cookson, Laudicella, & Donni, 2013). Another study from the English NHS examined the possible effects of increased financial incentives on equity and found that it could lead to reduction of inequalities in delivery of health care (Doran, Fullwood, Kontopantelis, & Reeves, 2008). In contrast to these re-sults, other studies of the English NHS have found that increased competition, similar to a patient choice system in primary care led to unclear effects on equity (Asaria, Cookson, Fleetcroft, & Ali, 2016; Goddard, Gravelle, Hole, & Marini, 2010). A few studies have examined how the Swedish PCCR has affected equity. These studies have mainly been performed in the county councils of Stock-holm and Skåne. The results show that the PCCR seem to have increased pa-tient visits for all groups (Agerholm, Bruce, Ponce de Leon, & Burström, 2015; Beckman & Anell, 2013). However, there seems to be some tendencies that the increase in patient visits have been smaller for patients with high

24

health care needs (Agerholm et al., 2015; Beckman & Anell, 2013). Taken together, the empirical studies that have examined what effect patient choice has on equity have inconclusive results that is difficult to interpret.

25

Methods and findings

This thesis builds on four studies that were all carried out in Sweden. The first study focuses on how local governments have tried to formulate contracts when contracting out residential elder care. The second and third study centres on primary care and how the Swedish Patient Choice reform has affected eq-uity. The second study contains a descriptive quantitative analysis to assess the effect the Patient choice reform had on geographical equity based on the location of new primary health care centres (PHCCs). In the third study, the effect on equity was investigated by looking at what types of patients have been registered with private and public PHCCs. Lastly, the fourth study fo-cuses on how local governments try to counteract risk selection, or cream-skimming, by the design of local reimbursement systems. The four studies address different aspects of how local authorities can design and steer markets. By looking at both primary care and elder care, a broader understanding of how local authorities can design markets to reach public goals can be reached.

Study I – Contracting out care homes The aim of this study was to examine how local governments in Sweden write quality requirements when contracting-out residential elder care. In the analysis, we examined contracts from four municipalities that started to contract-out residential eldercare immediately after the Swedish legislation change in 1992 and which all have a relatively large share of their residential homes run by private contractors (between 20-80%). In this sense, the selected municipalities can be considered as ‘most likely’ cases; they have a relatively long and extensive experience of contracting in this area. The ‘most likely’ case selection logic implies that if outcomes are poor, it is likely that other municipalities with less favourable circumstances will perform as bad or worse (King, Keohane, & Verba, 1994).

Each contract contains around 100-200 so called quality requirements where the local government specify what the provider should provide. In total 1005 quality requirements were analysed. In order to assess the ability to monitor the quality requirements, these were classified into three different categories

26

depending on how monitorable they were perceived to be. The first category consists of those requirements which cannot be monitored. They are formu-lated in such a loose and general fashion that it is impossible to determine whether or not they have been met. This category was labelled “non-mon-itorable.” The second category contains requirements formulated in such a way that they could be monitored by giving a ‘yes’ or ‘no’ answer as to whether or not they have been fulfilled. In this sense, the requirements in this category are possible to monitor in a dichotomous way but are still non-meas-urable in a quantitative way. Such requirements are labelled ‘partly mon-itorable’ as we see them as monitorable in a more crude sense. The third cat-egory consists of requirements which we see as fully monitorable; that is, when it is possible to determine also the extent to which a quality requirement formulated in the contract has been met. These requirements are thus formu-lated in a way that makes them measurable and quantifiable; this implies that it is possible to determine the degree to which the requirement is fulfilled by the provider, and if not, how far actual quality standards are from the target. This provides considerably more information both about the quality objectives of the principals, e.g., what this actor really wants, and the quality of the ser-vice performed.

Results Of the quality requirements analysed there was a clear focus on care work and social activities, organisation and staffing as well as rehabilitation. Most of the requirements were classified as party monitorable (71%) and only a small fraction of the requirements were classified as fully monitorable, i.e. both monitorable and measureable (5%). The share of the quality requirements that were classified as not at all monitorable was 25 percent.

A majority of the quality requirements concerned processes (80%) whereas almost 20 percent of the quality requirements concerned structural quality. Outcome requirements were very rare, less than 1%.

The results showed that the ‘softer’ areas of care work and social activities to a considerably less amount could be considered to be monitorable than ‘harder’ areas such as law, taxes and economy, and organisation and staffing. This indicates that the risk of moral hazard could be higher when contracting-out ‘soft services’ compared to ‘hard’ services such as garbage disposal.

27

Study II – How does patient choice affect geographical equity The aim of this study was to examine what effect the patient choice reform in primary care has had on the geographical distribution of health care services. What types of providers (public or private) established PHCC’s after the in-troduction of the patient choice reform and where were they located? To in-vestigate this, a cross-sectional data set containing socio-economic data of the geographical areas of all primary health care centres in Sweden was used.

The study was carried out by analysing where new PHCC had established themselves in the 21 county councils. The design of the study was to compare socio-economic data on individuals who reside in the same electoral areas in which the primary health care centres in Sweden are located. The socio-eco-nomic data on the persons living in these areas was gathered from Statistics Sweden and consists of the following variables:

Mean income Percentage of citizens born outside western Europe and North

America Percentage of single mothers/fathers living with children under 18

years Percentage of low-educated Percentage of unemployed or in public programs Percentage of people over 65 years living alone Percentage younger than 5 years

Since all primary health care centres belong to a specific county council and thereby are affected by different reimbursement systems as well as different county characteristics, the primary health care centres were grouped in a sec-ondary level variable, i.e. the county councils.

To examine whether there are any differences in regard to where old and new primary health care centres choose to locate after the patient choice reform, 2-sided t-tests were used. In our main models where we controlled for county councils and municipalities, generalized estimating equations (GEE) models were used. This method was used since the socio-economic variables can be correlated within certain municipalities and/or county councils and regular OLS regressions therefore would risk violating the independence assumption (Hubbard et al., 2010).

28

Results Of the 285 PHCC’s that were established after the introduction of the PCCR, 270 (95%) was owned by private companies. This number can be compared to the situation before the reform where only 30% of the PHCCs were pri-vately owned. When comparing the areas where old and new PHCCs are located, the results showed that new PHCCs were located in areas with statistically significantly fewer single parents, fewer older people living alone, higher mean income, lower degree of unemployment and fewer low-educated people. When analysing the data using GEE-models to control for the effect of the county councils, the results show a significant effect on three variables: new PHCCs had fewer people over 65 years living alone, fewer single parents, and a higher percentage of immigrants from outside of Western Europe. However, when including multiple socio-economic variables in the same model and choosing the model with the best fit according to the QICC-score, there were no statistically significant effect on percentage of immigrants outside of West-ern Europe. Still, the significant effect of people over 65 years living alone and single parents remained. This indicates that new private providers do have a tendency to establish in areas that are relatively affluent.

Study III – Risk selection in Swedish primary care The aim of this study was to assess socio-economic differences between pa-tients registered with private and public primary health care centres. The de-sign was a population-based cross-sectional study controlling for municipality and household. Socioeconomic data on all individuals residing in Sweden (n= 9,851,017) was collected from Statistics Sweden and linked to individual reg-istration data from all 21 Swedish regions. The data set thereby makes it pos-sible to identify which PHCC each Swedish individual is registered with. The main outcome variable was ownership of the PHCC an individual was registered with. Socioeconomic individual level variables were selected based on previous research suggesting an association between the variable and dif-ferences in health needs and. The following socioeconomic variables were in-cluded in the analysis:

Country of birth (grouped into clusters) Disposable yearly income per household member in SEK Highest level of finished education Recipient of sickness benefits

29

Recipient of unemployment compensation Recipient of labour market compensation Recipient of early retirement compensation

A linear probability model was used to analyse the probability of selecting private PHCC conditional on socio-economic variables. Apart from analysing all individuals in a total population-model, two different samples were sepa-rately analysed. The samples included only individuals living in municipali-ties that could be considered to have access to alternative PHCCs.

Results After controlling for municipality and household, individuals with higher so-cio-economic status were more likely to be registered with a private primary health care provider. Individuals in the highest income quantile were 4.9 per-centage points (13.7%) more likely to be registered with a private primary health care provider compared to individuals in the lowest income quantile. Individuals with one to three years of higher education were 4.7 percentage points more likely to be registered with a private primary health care provider compared to those with an incomplete primary education. Individuals born in Africa and Asia were respectively 5.6 and 3.5 percentage points less likely to be registered with a private PHCC compared to individuals born in Sweden. The results show that there are notable differences in registration patterns, in-dicating a skewed distribution of patients and health risks between private and public primary health care providers.

Study IV – Risk adjustment measures in Swedish primary care In this study, the aim was to analyse how the county councils, have acted in order to combat risk selection in primary care following the introduction of the primary care choice reform. To answer the question, we analysed contracts and reimbursement systems from all county councils in two different years (2013 and 2016). A framework for analysis was constructed and the material was classified deductively. Furthermore, data on political majorities in the counties at the two time points were gathered.

Results The results showed that the counties have used a multitude of strategies to combat risk selection, such as financial risk adjustment and requirements re-garding the scope and content of the provided services. Furthermore, it seems

30

as if though ideology was associated with the degree of risk adjustment in 2013, i.e. county councils with left-wing majorities were more likely to com-bat risk election to a higher extent. However, the difference between counties governed by right-wing and left-wing majorities disappeared in 2016 where a convergence between the counties was noticed. These results imply that a pol-icy learning process has taken place where counties use empirical data on how their system works and get ideas from other counties on how to combat risk selection, regardless of ideology.

31

Discussion

Principal findings When local governments create quasi-markets they have two main ways of steering the providers: by specifying what should be done through contracts or by creating reimbursement systems that incentivize desirable behaviour from the providers. The studies included in this dissertation show that both ways of steering comes with difficulties. A general finding is that it is difficult for local governments to specify what quality is in soft services such as elder care. Furthermore, despite several efforts from county councils to increase eq-uity in access to primary care, there still seem to be equity issues in regards to access to care. The results from study I indicate a clash between the logic of competitive con-tracting and the logic of achieving quality in eldercare. Since competition be-tween different providers in a competitive contracting type of public procure-ment is undertaken ex-ante, it is the principal, in this case the municipalities, which assess the quality of the different tenders before the provision of the services has started. What constitutes quality of elder care is to a large degree subjective, i.e. different users of elder care services have different needs and desires. The principals have the intricate task of assessing quality of care for all individuals that receive elder care whilst they must do so in a neutral way. Furthermore, in line with the results from Almqvist (2001) and Almqvist and Högberg (2005), a large majority of the quality requirements were focused on processes and structures rather than outcomes or results. This implies that con-tracting municipalities try to control the process, e.g. how tasks are carried out, rather than the end result. This could arguably present a problem accord-ing to contracting theory (T. L. Brown, Potoski, & Van Slyke, 2009; O Hart, Shleifer, & Vishny, 1997). The basic premise when a principal, e.g. a munic-ipality, contracts-out a service to an agent is a belief that the agent will be able to achieve similar or better output to a lower price (M. A. Bergman, Johansson, Lundberg, & Spagnolo, 2016; Hefetz, Warner, & Vigoda-Gadot, 2014). If the principal stipulates exactly how an agent should act by formulat-ing requirements focusing primarily on processes and structure, the agent has its hands tied and limits the possibility to produce the services in a more effi-cient and innovative way.

32

The results from study II show that there seem to be some negative effects on geographical equity after the introduction of the patient choice reforms. New providers have mainly established themselves in areas with higher socio-eco-nomic status – a pattern that may lead to a skewed allocation of health care resources. However, the effects are not very large and also seem to vary be-tween county councils. A possible explanation for this could be the construc-tion and design of the reimbursement systems which vary between all county councils and could buffer some of the unintended effects of the patient choice system. For instance, some county councils have chosen to weigh the capita-tion reimbursement based on socio-economic status or diagnoses among the listed patients. The incentive structures are therefore significantly dissimilar in different county councils. Study III strengthens the results from study II and shows that there are sub-stantial socio-economic differences between individuals registered with pri-vate and public PHCCs. When controlling for municipality we can still find that individuals registered with private PHCCs have higher income, higher education and are more likely to be born in Sweden compared to individuals registered with a public PHCC. These results indicate that we might see indi-cation of cream skimming in the Swedish primary care where private PHCCs try to attract more profitable individuals to register with them. This can be done both by locating in more affluent areas within a municipality but also through other indirect means such as offering a certain set of services that is likely to attract patient groups with low costs. An alternative explanation to the results found in study III is that individuals with higher levels of income and education are more prone to choosing a pri-vate PHCC. Highly educated individuals are generally more likely to make active choices of providers and therefore can be expected to choose private PHCCs to a higher degree (Dixon, Robertson, Appleby, et al., 2010). Alterna-tively, individuals with a high socio-economic status could be more ideologi-cally keen to choose private PHCCs since high socio-economic status are cor-related with having more liberal ideological viewpoints (Brooks, Nieuwbeerta, & Manza, 2006; Elff, 2009). Taken together, it is difficult to assess to what degree differences in socio-economy between individuals reg-istered with private and public PHCCs can be explained by mechanisms at the provider level or individual level. I argue, based on the results from study II and study III as well as previous research that it is likely to be a combination of both. Regardless of the cause of socioeconomic differences in PHCC reg-istration patterns, it can still be seen as a problem from an equity perspective. If individuals that are less likely to have high health care needs are more likely to register with private PHCCs, this can lead to a skewness in the resource distribution for primary care. If individuals with high socio-economic status and low health care needs are registered with a PHCC in a county council, this

33

means fewer resources for other PHCCs in that county council where individ-uals with high health care needs are registered. If these PHCCs are publicly owned, it is possible for the county council to share the costs between them, but this is not viable if the PHCCs are owned by different legal entities.

In an attempt to combat this potential skewness, the county councils can use a range of different strategies. To what extent different strategies have been em-ployed and how the adaptation of these policies have developed over time was examined in study IV. The results showed that the counties have used a mul-titude of strategies to try to combat risk selection. Furthermore, it seems as if ideology could explain the degree of risk adjustment at the onset of the intro-duction of the patient choice-systems where left-wing majorities were more likely to adopt several methods to combat risk selection. However, the differ-ence between counties governed by right-wing and left-wing majorities disap-peared later on, where we could see a convergence between the counties. This result implies a policy learning process where counties use empirical data on how their system works and get ideas from other counties on how to combat risk selection. Additionally, it indicates that the political goals relating to ac-cess in health care does not differ much between left-wing and right-wing ma-jorities.

Methodological considerations When considering how to conduct sampling in the different studies we have tried to be thorough and use a sensible and purposeful sampling strategy. In Study I, a most-likely case sampling was utilised for the purpose of creating more ‘generalizable’ or ‘transferable’ conclusions. In Study II, III and IV, all possible units of analysis were included in the sample, thus minimising the problems associated with sampling. However, in study III we decided to, apart from the total population, analyse the data with samples where individuals living in municipalities without alternative PHCCs to choose from were ex-cluded. In study IV, we chose to include data from only two years. The reason behind this is that we wanted to capture changes in the political majorities in the county councils which regularly only happens each fourth year.

When performing a regular OLS regression one of the main assumptions is independence between the different units of analysis. E.g. if we want to study the effect on a certain treatment on patients and these patients are being treated at different clinics, the possibility exist that the effect of the treatment is de-pendent upon which clinic a patient belongs to, i.e. the data is clustered. In the case of study II and study III, the unit of analysis are all Swedish PHCC that all belong to different clusters in the form of municipalities and county coun-cils. The municipalities could have a clustering effect since a municipality

34

could have a higher mean income than its neighbouring municipality or in other ways have different socio-economic characteristics. The PHCCs are also clustered within a county council that can affect the results since each county council has a certain reimbursement system and other regulations that affect all PHCCs in that county council. Thus, it is important when working with clustered data to somehow control for this cluster effects (Begg & Parides, 2003). When doing this there are several possibilities, e.g. one could control for the county councils using a fixed effect model, a random effect model or, as we opted for, utilise a generalized estimating equation modelling (GEE) technique. In Study II we opted for a GEE approach since we are not interested in the effects of belonging to a certain municipality or county council but ra-ther treats it as a nuisance that we wish to control for (Begg & Parides, 2003; Gardiner, Luo, & Roman, 2009). In study III, data from all individuals residing in Sweden was gathered. Due to the large number of units of analysis, the issue with clustering that was noted above could be solved by controlling for dummy-coded municipalities in a fixed-effect model. A possible solution would have been to instead utilize a random-effect model but since the fixed-effect model requires fewer as-sumptions (e.g. no need to make assumptions regarding the error of the ran-dom effect) we considered it to be a better choice for these analyses. Another methodological consideration that arose in both Study II and Study III is whether it is suitable to control for any variables in the regression mod-els. One could argue here that full multiple regression models are more inter-esting since we, when controlling for other variables get closer to the ‘true effect’ of a single predictor variable. However, I believe that in both Study II and Study III, it is of more value to look at the crude estimates where we only control for geographical areas. Since we are not interested in developing a causal or predictive model but rather use the socioeconomic variables of in-terest as a proxy for socio-economy, I argue that the crude estimates with no control variables added give a better indication of socio-economic differences between private and public PHCCs. In the studies, we chose to include several, both crude and full, models, which allows the readers to decide for themselves what models are of most interest. An issue that was noticeable in both study II and study III is how to operation-alise and measure socio-economy (SES). This is a subject that have been dis-cussed in length by previous research (P. A. Braveman et al., 2005; Ghawi et al., 2015; Rubin et al., 2014). It is difficult to find a clear consensus regarding how to define and operationalize socio-economy but in general SES can be said to consist of a combined measure of several variables such as an individ-ual’s income, level of education and social status (P. A. Braveman et al., 2005; Marshall et al., 2015). Some scholars have argued that objective measures of

35

SES such as income and level of education overlook intersectional aspects of SES that can be better captured by combining objective measures with self-evaluations (Rubin et al., 2014). However, in the studies included in this dis-sertation, we have been limited to measures that can be gathered from Statis-tics Sweden for all individuals residing permanently in Sweden. This limita-tion excludes the use of more subjective measures of socio-economy such as self-identification of SES. An alternative approach in both Study II and Study III would have been to create indices for socioeconomic status by adding mul-tiple variables together. One advantage with this approach would have been that it might have been easier to interpret for the reader. A disadvantage how-ever, is that we would lose information and detail since the association be-tween single socio-economic measures and the outcome would become hid-den. In both Study I and IV, a question that arose was how to deal with quantitative data in studies that had lot of qualitative elements. Especially Study I with an inductive coding process, was to some degree subjective, meaning that a dif-ferent coder probably would not have coded all quality requirements exactly the same way. Due to the approximate nature of the data in these studies, we argue that formal statistical tests such as t-tests were not suitable. A formal statistical test would give the reader a notion of exactness that did not correlate with the validity of the coding process. Instead, we chose to use simple de-scriptive statistics to show differences between groups and complement this with more examples of how different items were coded. In this way, transpar-ency increases and the readers can themselves follow the process and decide if our conclusions are valid.

36

Theoretical and policy implications

The studies carried out in this dissertation indicate that there might be a con-flict between quasi-market models as they have been introduced in Sweden and the ability for local governments to design markets so as to achieve equity and quality in health and social services. These results indicate that the theo-retical approach to contracting needs to complemented by looking further into the monitorability of contracts. Further, more emphasis should be put on the possibility to steer markets depending on what type of contracting mecha-nisms are being used. Competitive contracting via public procurement creates different opportunities to monitor an agent since the competition between ten-ders occurs before the services are being performed compared to the case in a user choice system where the competition occurs during and after the services are being provided. Thus, when using competitive contracting mechanisms to contract out, it creates a situation where the principal needs to specify all re-quirements in advance which, as shown in study I, can be problematic. When a principal utilizes a user choice system to contract-out other obstacles ensue. Compared to competitive contracting the principal loses control in a user choice system with free establishment for providers. This means that the prin-cipal no longer has the ability to control where providers are established and these decisions are instead transferred to the quasi-markets which consist of the providers and the user/patients. To be able to achieve desirable goals in a user choice system, the principal must take a more reclined approach and try steering through the creation of different incentive structures. Is it possible for local authorities to design patient choice systems in a way to minimizes equity issues? As shown in the studies in this dissertation, this is difficult. Within primary care, there is a distinct imbalance in where new PHCCs have been set up and what individuals are registered with private PHCCs. The differences are systematic and consistent over the whole country. The central government and the county councils have tried to mediate this by forbidding a PHCC to refuse an individual to register and to risk adjust the reimbursement based on socioeconomic variables, diagnoses and distance to a large city. Despite these efforts, we still see tendencies towards risk selec-tion. This implies that the steering and design of the markets does not work completely as intended.

37

A possible remedy is to increase the degree of risk adjustment even more since there is clear evidence that providers respond to financial incentives (Andersen, 2009; Anell, Dackehag, & Dietrichson, 2018; Barros, 2003; Croxson, 2001). Thus, it is reasonable to believe that an increase in the degree of risk adjustment would affect the behavior of the providers such that they try to locate in less socioeconomically affluent areas and attract patient groups with higher health care needs. However, this solution might not be entirely satisfactory. First, there is the question what type of risk adjustment measure to use. Today in the Swedish PCCR, there are mainly two types of measures that are used to adjust the capitation reimbursement: Care Need Index (CNI) and Adjusted Clinical Groups (ACG). CNI is beneficial mainly since it is not possible to affect for the providers. It is calculated using official statistics of aspects such as age and country of birth of the enlisted patients. This means that it is inex-pensive to use and the local authorities do not need to be concerned about providers manipulating the data. However, if the goal of using CNI is to cap-ture health care needs, it is not the best measure to use. Since CNI only cap-tures socioeconomic variables it can only be used indirectly to measure health care needs. It is for instance reasonable to believe that there are large varia-tions in health care needs among different 75 year old people that live alone (one of the variables included in CNI). Furthermore, if the county councils instead put more emphasis on diagnoses-based risk adjustment such as ACG, the advantage is that it measures health care needs more directly. The disad-vantage with using ACG is that it is based on diagnoses, meaning that provid-ers can game the system by recording more severe diagnoses or irrelevant di-agnoses for patients, thereby increasing their revenue. The county councils can monitor this by scrutinizing journals to see if the recorded diagnoses are correct and reasonable. This is however costly to do and it can be seen as a type of transactional costs for contracting-out health care. Furthermore, there is also a risk for increased costs for the providers since they need to make sure all diagnoses are recorded correctly at each visit. Some county councils have worked around this problem by only calculating ACG using diagnoses rec-orded by specialists at hospitals. This could be a solution but means the ACG-scores are not as correct since all diagnoses are not included. Second, increasing the degree of risk adjustment does not solve an intrinsic problem with adjusting reimbursement based on characteristics of the regis-tered patients. Since all listed patients on a PHCC contribute to the capitation payment and the capitation is based on a computed average costs to treat a single patient over a year, there will still be individuals, i.e. patients with the highest health care needs, that will not be profitable for the providers. Irre-spective of the degree of risk-adjustment, the most profitable patients in a re-imbursement system that relies on capitation will always be the patients that

38

do not visit the PHCC. Conversely, the most costly patients are those with the most visits to a PHCC. This creates incentives for the providers to not be as responsive to the needs of these groups of high costs patients and indirectly push them to choose a different provider. This is in direct contrast to the main principles behind a patient choice system – that the providers should compete with high quality to try to attract patients. To remedy this, a solution that have been put forward in other health care systems, is for the health care authorities to create a separate high cost pool for patients that are not profitable, i.e. pro-viders are reimbursed from a separate pool of money, for the small group of patients with the highest costs (Schillo, Lux, Wasem, & Buchner, 2016). Third, despite most county councils having tried to increase the reimburse-ment for PHCCs outside large cities, there are still very few PHCCs that have been established in areas that are more rural. It seems difficult to incorporate a reasonable reimbursement policy that can work for both urban and rural set-tings. A problem is the large scope of services that many county councils de-mands all PHCCs to provide, which makes it difficult to establish a PHCC in areas with fewer potential patients. A possible solution could be to create dif-ferent patient choice systems for urban areas and rural areas that can take into consideration the special challenges that are present when creating a reim-bursement model in different areas. Taken together, the studies presented in this dissertation have clear implica-tions for policy, not least in Sweden. Given that policy makers both at the national and local levels still make equity a high priority within the system, it seems as if there is a need to review to the legal and financial conditions for establishment of new private care providers. Furthermore, national and local authorities should pay close attention to how contracts with private providers of health and elder care are written when contracting-out services to these, not least with regards to the possibility for the local authorities to monitor out-contracted services.

39

Further studies

The conclusions of this dissertation call for further research in several areas. Among these, a better understanding of how reimbursement systems and risk adjustments affect listing patterns within a patient choice system. This could be done by utilizing individual level longitudinal data and study changes in reimbursement system and the effect this has on what PHCCs patients are reg-istered with. Furthermore, studies are needed that look into how local governments can uti-lize combined mechanisms when contracting out and how these can work to-gether. For instance, it is possible to combine ex-ante competition with quality requirements with ex-post competition through for instance a patient or user choice system. What incentives can these combinations create and what are the effects on quality of care and equal access to care?

40

Summary

Municipalities and county councils use a multitude of different strategies when they design and steer health and social care markets to try to meet goals such as quality and equity. Depending on the strategies used, different prob-lems arise. The aim of this thesis is to examine how local authorities can de-sign quasi-markets in a way that achieves public goals such as equity and high quality. To answer the aim, four empirical studies were carried out.

The studies show that when designing a market by contracting-out through public procurement, the issues lay primarily at specifying and defining what is meant by quality before a service is privatized. This is especially difficult to do concerning soft areas such as elder- and healthcare. If this is not done properly, it can lead to crucial issues for monitoring quality since the contract-ing authority cannot hold the provider responsible for delivering an aspect of a service if that aspect is not specified in the contract.

When a market is designed as in the patient choice systems in primary care, it creates a whole other set of difficulties for the local governments. Here, it is not as important to specify quality beforehand in the contracts since quality monitoring is done retrospectively by both the counties themselves as well as the patients who with their choices can monitor quality by punishing providers with poor quality by registering with another provider. Instead, the crucial problem is how to design reimbursement systems that will lead to equal access to health care. In this respect, the county councils utilize different methods. Despite these measures, the primary care choice reform has led to inequity, both geographically in regards to where new private primary health care cen-tres are located but also and to a larger degree, socio-economic inequity relat-ing to what kind of socio-economic groups of individuals are registered with private PHCCs. In other words, county councils do not manage to fully coun-teract risk selection behaviour by the design of their reimbursement system, which could imply issues with unequal access to health care.

41

Summary in Swedish

Svensk sjukvård och äldreomsorg bygger på två huvudprinciper: solidaritet och jämlikhet. Detta har tagit sig uttryck i att verksamheten i huvudsak är skattefinansierad skatter och fördelningen av resurser ska ske i enlighet med behovsprincipen. Under 1980- och 1990-talet infördes en omfattande moder-nisering av den offentliga sektorn i stora delar av världen, ofta benämnd New Public Management eller NPM. En av de mest påtagliga förändringar som kom från NPM-idéerna är privatiseringen av produktionen av välfärdstjänster såsom hälso- och sjukvård. Utvecklingen med NPM-inspirerade reformer såsom privatisering har varit tydlig också i Sverige. Sedan början av 1990-talet har vi i Sverige haft en kontinuerlig marknadsanpassning av välfärdssy-stemet där privata sjukvårds- och omsorgsföretag har bjudits in att och kon-kurrera med offentliga leverantörer. Vissa har dock menat att privata, vinst-drivande aktörer inom sjukvård och äldreomsorg kan innebära ett hot mot be-hovsprincipen inom sjukvården.

När man styr sjukvården måste politiker och byråkrater se till att lagar följs och att offentliga mål som jämlik tillgång till vård och högkvalitativ vård är uppfyllda. I en traditionell byråkratisk organisation kan politiker och tjänste-män mer direkt styra produktion och fördelning av välfärdstjänster. När ett landsting eller en kommun lägger ut driften av en tjänst på en privat utförare krävs andra verktyg för att kunna styra verksamheten. För att uppnå önskvärda mål som jämlik tillgång till högkvalitativ vård måste de offentliga myndighet-erna som ansvarar för sjukvård och äldreomsorg skapa rätt incitament för pri-vata vårdproducenter. Detta kan till exempel göras genom att specificera kva-litetskriterier som de privata leverantörerna måste följa för att uppfylla ett kon-trakt, eller genom att skapa ekonomiska incitament för utförarna. Att styra en upphandlad verksamhet skapar således nya utmaningar för kom-muner och landsting. De övergripande syftet med denna avhandling är att un-dersöka hur lokala myndigheter kan utforma kvasi-marknader så att politiska mål som hög kvalitet och jämlik tillgång till vård kan uppnås. I den första studien undersöktes hur kommuner skrev avtal med privata företag vid upphandling av äldreboenden och om kvalitetskraven i dessa avtal kunde sägas vara uppföljningsbara. En kvalitativ induktiv innehållsanalys gjordes för att analysera totalt åtta kontrakt från fyra olika kommuner. 1005 kvalitetskrav

42

identifierades och kodades utifrån vilket område de fokuserade på, huruvida kravet rörde strukturer, processer eller utfallskvalitet samt i vilken grad kvali-tetskravet var uppföljningsbart. Analysen visade att kvalitetskrav som rörde mjuka områden såsom sociala aktiviteter och omvårdnad hade en mycket låg grad av uppföljningsbarhet. Dessa krav karaktäriserades av att de var skrivna på ett vagt och oprecist sätt vilket innebär att det är svårt för kommunen att avgöra huruvida ett sådant krav är uppfyllt. I den andra studien var syftet att undersöka hur vårdvalsreformen i primärvår-den har påverkat den geografiska jämlikheten. Detta gjordes genom att under-söka i vilken typ av områden vårdcentraler som öppnade efter vårdvalets in-förande ligger och jämföra detta med de områden där tidigare etablerade vård-centraler ligger. Forskningsdesignen byggde på att analysera socioekono-miska uppgifter om personer som bor i samma valdistrikt där de 1411 primära vårdcentralerna i Sverige är etablerade. Resultaten visade att vårdcentraler som etablerats efter vårdvalets införande låg i områden med signifikant färre äldre ensamboende samt färre ensamstående föräldrar - grupper som i allmän-het har lägre socio-ekonomisk status och ett högt förväntat sjukvårdsbehov. Emellertid observerades inga signifikanta skillnader för andra socioekono-miska variabler som medelinkomst, andel invandrare, utbildningsnivå och ar-betslöshet. I den tredje studien var syftet att undersöka socioekonomiska skillnader mel-lan patienter registrerade hos privata och offentliga primärhälsovårdscen-traler. Designen var en befolkningsbaserad tvärsnittsstudie där vi kontrolle-rade för kommuntillhörighet. Socioekonomiska uppgifter om alla personer bo-satta i Sverige (n = 9 851 017) samlades in från SCB och sammanlänkades med uppgifter om vilken vårdcentral alla var listade på. Individer med högre socioekonomisk status var i högre utsträckning listade på privata vårdcen-traler. Exempelvis var individer i den högsta inkomstkvantilen 4,9 procenten-heter mer benägna att lista sig hos en privat primärvårdspersonal jämfört med individer i den lägsta inkomstkvantilen. Vuxna individer med en universitets-utbildning på 1-3 år var 4,7 procentenheter mer benägna att vara listade hos en privat vårdcentral jämfört med individer som gått mindre än 9 år i grund-skola. Resultaten visar att det finns tydliga skillnader i listningsmönster base-rat på socioekonomi.

I den fjärde och sista studien var syftet att analysera hur landstingen har agerat för att motverka att vårdcentraler försöker attrahera framförallt friska indivi-der med förväntat lågt vårdbehov att lista sig hos dem, så kallad riskselektion. Vidare ville vi ta reda på om politisk majoritet kunde förklara i vilken utsträck-ning landstingen försökt motverka riskselektion. För att undersöka detta ana-lyserade vi kontrakt och ersättningssystem från alla landsting för åren 2013 och 2016. Analysen visade att landstingen har använt ett flertal olika strategier

43

för att försöka bekämpa riskselektion, däribland olika former av justeringar av ersättningen. Vidare verkar det som att ideologi skulle kunna förklara i vilken grad landstingen försökte motverka riskselektion under 2013 då landsting som styrdes av en vänstermajoritet var mer benägna att använda flertalet åtgärder för att motverka riskselektion. Dock minskade dessa politiska skillnader år 2016, där vi istället kunde se en konvergens mellan landstingen. Detta resultat kan innebära att vi ser en politisk inlärningsprocess där landstigen använder empiriska uppgifter om hur deras styrning fungerar samt får idéer från andra län om hur man kan motverka riskselektion och på sätt lär sig både av sig själv och av varandra. Sammantaget visar avhandlingen att kommuner och landsting använder en mängd olika strategier när de utformar och styr marknader för sjukvård och äldreomsorg för att försöka säkerställa att verksamheten bedrivs med hög kva-litet och på ett jämlikt sätt. Beroende på vilka strategier som används uppstår dock olika problem. Vid offentliga upphandlingar är ett tydligt problem hur man ska specificera och definiera vad som menas med kvalitet innan en tjänst privatiseras. Det här är särskilt svårt att göra när det gäller mjuka områden som äldreomsorg. Om detta inte görs korrekt kan det leda till stora problem när en kommun eller landsting ska följa upp kvaliteten på tjänstens utförande då en kommun måste kunna peka på vilket kvalitetskrav som inte är uppfyllt om man är missnöjd med kvaliteten på den levererade tjänsten. När en marknad är utformad som en kvasi-marknad, såsom i vårdvalet i pri-märvården, skapar det en helt annan uppsättning svårigheter för de lokala myndigheterna. Här är det inte lika viktigt att specificera kvalitetskrav i kon-trakten eftersom kvalitetsövervakningen, om allting fungerar som det är tänkt, sker av patienterna eller brukarna som med sina val kan ”straffa” utförare med dålig kvalitet genom att välja en annan utförare. Istället är en av de mer grund-läggande utmaningarna i dessa system att utforma vårdvals- och ersättnings-system som leder till jämlik tillgång till vård. I detta avseende använder lands-tingen olika metoder. I avhandlingen visas emellertid att trots dessa åtgärder leder vårdvalet inom primärvården fortfarande till ojämlikhet. Både geogra-fisk ojämlikhet när det gäller var nya privata vårdcentraler etablerar sig, men också i större utsträckning när det gäller listning där individer med hög socio-ekonomisk status i högre grad är listade hos en privat vårdcentral. En slutsats som kan dras från denna avhandling är att landstingen inte fullt ut lyckats med att motverka riskselektion vilket riskerar att leda till problem med ojämlik till-gång till sjukvård.

44

Acknowledgements

Många personer har på olika sätt bidragit till att jag har kunnat färdigställa denna avhandling och jag tänkte ta tillfället i akt att tacka dessa. Min huvud-handledare Ulrika Winblad som på ett engagerat, tålmodigt och kunnigt sätt har lett mig genom den långa process som ett doktorandprojekt innebär. Du har alltid funnits tillgänglig för frågor kring stort som smått och agerat boll-plank när det behövts. Min bihandledare Paula Blomqvist som med ett skarpt sinne så tydligt kan lyfta upp empiriska studier till ett mer generellt och teore-tiskt plan. Ni båda har varit väldigt engagerade som handledare och på ett ge-neröst sätt delat med er av all er kunskap som jag efter bästa förmåga har för-sökt att ta till mig.

Vidare vill jag tacka alla medlemmar i gruppen för hälso- och sjukvårdsforsk-ning som på olika sätt bidragit till både en trivsam arbetsmiljö, skarpa diskuss-ioner kring design och teori samt insiktsfulla kommentarer på manuskript. Mio Fredriksson som för det första var den som introducerade mig till fors-kargruppen. Du har också varit en person som utöver att agera träningssällskap även varit en nära vän under min doktorandtid som alltid har haft tid över för att diskutera såväl forskning som annat roligt. Hoppas på fortsatt samarbete framöver! Tack också till min rumskamrat Jan Larsson som alltid har så kloka tankar kring det mesta i livet. Mina doktorandkollegor från gruppen som är på plats i korridoren så gott som dagligdags, Linda Moberg, Caroline Hoffstedt, Sofie Vengberg, Linn Kullberg, Anna Mankell, Helene von Granitz, Carina Ahlstedt och Alexander Tegelberg. Vi har haft mycket utbyte kring idéer om våra avhandlingar och jag hoppas på att detta kan fortsätta på andra sidan dis-putationen. Tack också till Fredrik Olsson, Anna Hallberg, Douglas Spangler och Ylva Lindberg med flera forskningsassistenter som bidragit till att skapa en kontinuitet och trivsam arbetsmiljö i forskargruppen. I forskargruppen vill jag också särskilt tacka alla seniora forskare som delar med er av sin visdom. Inger Holmström, Margareta Sanner, Elenor Kaminsky, Marta Röing, Tobias Dahlström, Ann-Catrine Eldh, Cecilia Bernsten, Inge-borg Björkman, Åsa Muntlin Athlin med flera.

45

Tack till alla som bidragit och fortsätter att bidra till att göra vår litteraturklubb till en sådan intressant institution. Vecka efter vecka bidrar alla till medryck-ande och lärorika diskussioner om vetenskapsteori, validitet eller hur man bör och inte bör skriva en vetenskaplig artikel. Det är alltid lika berikande att delta. På Institutionen för folkhälso- och vetenskap vill jag tacka alla som bidragit till inspirerande vetenskapliga samtal och en trivsam arbetsmiljö. Däribland Per Lytsy, Ingrid Demmelmaier, Martin Cernvall samt alla nuvarande dokto-rander. Också ett särskilt tack till Ronnie Pingel som tålmodigt har kommit upp med intressanta idéer på hur vi kan angripa diverse uppkomna statistiska problem.

Till sist vill jag tacka min familj. Mina föräldrar som från dag ett har inspirerat mig till att vilja lära mig mer. Martina som alltid stöttat mig, både genom kloka tankar kring vetenskap och sjukvård, men också genom att säga åt mig när det är dags att ta en liten paus från jobb. Tack också Martina för att du ställt upp så tålmodigt när jag arbetat många semesterdagar och helger under slutfasen med avhandlingen – du får tillbaka det om några år när det är din tur! Slutligen, tack till Tage och Stellan för att ni ger mig perspektiv på vad som egentligen är viktigt.

46

References

Agerholm, J., Bruce, D., Ponce de Leon, A., & Burström, B. (2015). Equity impact of a choice reform and change in reimbursement system in primary care in Stockholm County Council. BMC Health Services Research, 15(1), 420. http://doi.org/10.1186/s12913-015-1105-8

Almqvist, R. (2001). Management by Contract’: A Study of Programmatic and Technological Aspects. Public Administration, 79(3), 689–706. http://doi.org/10.1111/1467-9299.00275

Almqvist, R., & Högberg, O. (2005). Management by Contract - vad hände? Kommunal Ekonomi Och Politik, 9(2), 7–36.

Amirkhanyan, A. A., Kim, H. J., & Lambright, K. T. (2011). Closer Than “Arms Length”: Understanding the Factors Associated With Collaborative Contracting. The American Review of Public Administration, 42(3), 341–366. http://doi.org/10.1177/0275074011402319

Andersen, L. B. (2009). What determines the behaviour and performance of health professionals? Public service motivation, professional norms and/or economic incentives. International Review of Administrative Sciences, 75(1), 79–97. http://doi.org/10.1177/0020852308099507

Andersson, F., Janlöv, N., & Rehnberg, C. (2014). Hälso- och sjukvårdstjänster i privat regi. Konkurrens, kontrakt och kvalitet - hälso- och sjukvård i privat regi. Stockholm: ESO.

Anell, A. (2005). Swedish healthcare under pressure. Health Economics, 14(Suppl 1), S237-54. http://doi.org/10.1002/hec.1039

Anell, A., Dackehag, M., & Dietrichson, J. (2018). Does risk-adjusted payment influence primary care providers’ decision on where to set up practices? BMC Health Services Research, 18(1), 1–12. http://doi.org/10.1186/s12913-018-2983-3

Anell, A., Glenngård, A. H., & Merkur, S. (2012). Health Systems in Transition - Sweden health system review. Health Systems in Transition, 14(5), 187. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/22894859

Appleby, J., Harrison, A., & Devlin, N. (n.d.). What is the real cost of patient choice ? Consumer Policy Review, 15(3). Retrieved from http://web.ebscohost.com/ehost/pdfviewer/pdfviewer?sid=d4f169e8-7d02-4b28-9455-c9f4aa492ccf%40sessionmgr4&vid=4&hid=19

Arrow, K. (1985). The Economics of the Agency. In J. W. Pratt & R. J. Zeckhauser (Eds.), Principals and Agents: The Structure of Business (pp. 1–35). Boston: Harvard Business School Press.

Artz, K. W., & Brush, T. H. (2000). Asset specificity, uncertainty and relational norms: an examination of coordination costs in collaborative strategic alliances. Journal of Economic Behavior & Organization, 41(4), 337–362. http://doi.org/10.1016/S0167-2681(99)00080-3

47

Asaria, M., Cookson, R., Fleetcroft, R., & Ali, S. (2016). Unequal socioeconomic distribution of the primary care workforce: Whole-population small area longitudinal study. BMJ Open, 6(1), 1–9. http://doi.org/10.1136/bmjopen-2015-008783

Barr, D. a, Fenton, L., & Blane, D. (2008). The claim for patient choice and equity. Journal of Medical Ethics, 34(4), 271–4. http://doi.org/10.1136/jme.2006.019570

Barros, P. P. (2003). Cream-skimming, incentives for efficiency and payment system. Journal of Health Economics, 22(3), 419–43. http://doi.org/10.1016/S0167-6296(02)00119-4

Beckman, A., & Anell, A. (2013). Changes in health care utilisation following a reform involving choice and privatisation in Swedish primary care: a five-year follow-up of GP-visits. BMC Health Services Research, 13(1), 452. Retrieved from http://www.biomedcentral.com/1472-6963/13/452

Begg, M. D., & Parides, M. K. (2003). Separation of individual-level and cluster-level covariate effects in regression analysis of correlated data. Statistics in Medicine, 22(August 2002), 2591–2602. http://doi.org/10.1002/sim.1524

Bergman, M. A., Johansson, P., Lundberg, S., & Spagnolo, G. (2016). Privatization and quality: Evidence from elderly care in Sweden. Journal of Health Economics, 49, 109–119. http://doi.org/10.1016/j.jhealeco.2016.06.010

Bergman, M., & Lundberg, S. (2013). Tender evaluation and supplier selection methods in public procurement. Journal of Purchasing and Supply Management, 19(2), 73–83. http://doi.org/10.1016/j.pursup.2013.02.003

Bevan, G., Helderman, J.-K., & Wilsford, D. (2010). Changing choices in health care: implications for equity, efficiency and cost. Health Economics, Policy, and Law, 5(3), 251–67. http://doi.org/10.1017/S1744133110000022

Bevan, G., & van de Ven, W. P. M. M. (2010). Choice of providers and mutual healthcare purchasers: can the English National Health Service learn from the Dutch reforms? Health Economics, Policy, and Law, 5(3), 343–63. http://doi.org/10.1017/S1744133110000071

Bianchi, T., & Guidi, V. (2010). The comparative survey on the national public procurement systems across the ppn. Roma.

Blomqvist, P. (2004). The Choice Revolution: Privatization of Swedish Welfare Services in the 1990s. Social Policy and Administration, 38(2), 139–155. http://doi.org/10.1111/j.1467-9515.2004.00382.x

Blomqvist, P. (2005). The Turn to Privatization in Swedish Welfare Services : A Matter of Ideas ? Social Policy.

Boyne, G. A. (1998). Competitive tendering in local government: A Review of Theory and Evidence. Public Administration, Vol. 76, 695–712.

Braveman, P. A., Cubbin, C., Egerter, S., Chideya, S., Marchi, K. S., Metzler, M., & Posner, S. (2005). Socioeconomic Status in Health Research. Jama, 294(22), 2879. http://doi.org/10.1001/jama.294.22.2879

Braveman, P., & Gruskin, S. (2003). Defining equity in health 10.1136/jech.57.4.254. Journal of Epidemiology and Community Health, 57(4), 254–258. Retrieved from http://jech.bmj.com/content/57/4/254.abstract

Brooks, C., Nieuwbeerta, P., & Manza, J. (2006). Cleavage-based voting behavior in cross-national perspective: Evidence from six postwar democracies. Social Science Research, 35(1), 88–128. http://doi.org/10.1016/j.ssresearch.2004.06.005

Brown, T. L., Potoski, M., & Van Slyke, D. M. (2006). Managing Public Service Contracts: Aligning Values, Institutions, and Markets. Public Administration Review, 66(3), 323–331. http://doi.org/10.1111/j.1540-6210.2006.00590.x

48

Brown, T. L., Potoski, M., & Van Slyke, D. M. (2007). Trust and contract completeness in the public sector. Local Government Studies, 33(4), 607–623. http://doi.org/10.1080/03003930701417650

Brown, T. L., Potoski, M., & Van Slyke, D. M. (2009). Contracting for Complex Products. Journal of Public Administration Research and Theory, 20(Supplement 1), i41–i58. http://doi.org/10.1093/jopart/mup034

Burstrom, B. (2010). Looking to Europe - Will Swedish healthcare reforms affect equity ? BMJ, 340(January).

Cheng, S.-H. (2004). Physician performance information and consumer choice: a survey of subjects with the freedom to choose between doctors. Quality and Safety in Health Care, 13(2), 98–101. http://doi.org/10.1136/qshc.2003.006981

Cookson, R., Laudicella, M., & Donni, P. L. (2013). Does hospital competition harm equity? Evidence from the English National Health Service. Journal of Health Economics, 32(2), 410–422. http://doi.org/10.1016/j.jhealeco.2012.11.009

Croxson, B. (2001). Do doctors respond to financial incentives? UK family doctors and the GP fundholder scheme. Journal of Public Economics, 79(2), 375–398. http://doi.org/10.1016/S0047-2727(00)00074-8

Dahlgren, G. (2008). Neoliberal Reforms in Swedish Primary Health Care: For Whom and for What Purpose? International Journal of Health Services, 38(4), 697–715. http://doi.org/10.2190/HS.38.4.g

Dixon, A., & Le Grand, J. (2006). Is greater patient choice consistent with equity? The case of the English NHS. Journal of Health Services Research & Policy, 11(3), 162–6. http://doi.org/10.1258/135581906777641668

Dixon, A., Robertson, R., Appleby, J., Burge, P., Devlin, N., & Magee, H. (2010). Patient choice - How patients choose and how providers respond. London.

Dixon, A., Robertson, R., & Bal, R. (2010). The experience of implementing choice at point of referral: a comparison of the Netherlands and England. Health Economics, Policy, and Law, 5(3), 295–317. http://doi.org/10.1017/S1744133110000058

Domberger, S., & Jensen, P. (1997). Contracting out by the public sector: theory, evidence, prospects. Oxford Review of Economic Policy, 13(4), 67–78. Retrieved from http://oxrep.oxfordjournals.org/content/13/4/67.short

Donahue, J. D. (1989). The privatization decision : public ends, private means. New York: Basic Books. Retrieved from http://libris.kb.se/bib/4962342

Doran, T., Fullwood, C., Kontopantelis, E., & Reeves, D. (2008). Effect of financial incentives on inequalities in the delivery of primary clinical care in England: analysis of clinical activity indicators for the quality and outcomes framework. The Lancet, 372(9640), 728–736. http://doi.org/10.1016/S0140-6736(08)61123-X

Eisenhardt, K. (1989). Agency theory: An assessment and review. Academy of Management Review, 14(1). Retrieved from http://amr.aom.org/content/14/1/57.short

Elff, M. (2009). Social divisions, party positions, and electoral behaviour. Electoral Studies, 28(2), 297–308. http://doi.org/10.1016/j.electstud.2009.02.002

Ellis, R. (1998). Creaming, skimping and dumping: provider competition on the intensive and extensive margins. Journal of Health Economics, 17, 537–555. Retrieved from http://www.sciencedirect.com/science/article/pii/S0167629697000428

Elton, L. (2004). Goodharts Law and performance indicators in higher education., (May 2015), 37–41. http://doi.org/10.1080/09500790408668312

Forsberg, N. (2004). Offentlig upphandling i praktiken (3., [uppda). Stockholm: Norstedts juridik. Retrieved from http://libris.kb.se/bib/9619670

49

Forssell, A., & Norén, L. (2013). Kunden i kundvalsmodellen. Om konstitueringen av en kund i primärvården. Scandinavian Journal of Public Administration, 16(2). Retrieved from http://130.241.16.45/ojs/index.php/sjpa/article/view/1677

Fotaki, M. (2007). Patient Choice in Healthcare in England and Sweden: From Quasi-Market and Back To Market? a Comparative Analysis of Failure in Unlearning. Public Administration, 85(4), 1059–1075. http://doi.org/10.1111/j.1467-9299.2007.00682.x

Fotaki, M. (2010). Patient choice and equity in the British National Health Service: towards developing an alternative framework. Sociology of Health & Illness, 32(6), 898–913. http://doi.org/10.1111/j.1467-9566.2010.01254.x

Fotaki, M., Boyd, A., Smith, L., Mcdonald, R., Roland, M., Sheaff, R., … Elwyn, G. (2005). Patient Choice and the Organisation and Delivery of Health Services : Scoping review.

Fredriksson, M., Blomqvist, P., & Winblad, U. (2013). The trade-off between choice and equity: Swedish policymakers’ arguments when introducing patient choice. Journal of European Social Policy, 23(2), 192–209.

http://doi.org/10.1177/0958928712463158 Friedman, M., & Friedman, R. (1980). Free to choose. Gardiner, J. C., Luo, Z., & Roman, L. A. (2009). Fixed effects, random effects and

GEE: What are the differences? Statistics in Medicine, 28(2), 221–239. http://doi.org/10.1002/sim.3478

Ghawi, H., Crowson, C. S., Rand-Weaver, J., Krusemark, E., Gabriel, S. E., & Juhn, Y. J. (2015). A novel measure of socioeconomic status using individual housing data to assess the association of SES with rheumatoid arthritis and its mortality: a population-based case-control study. BMJ Open, 5(4), e006469. http://doi.org/10.1136/bmjopen-2014-006469

Goddard, M., Gravelle, H., Hole, A., & Marini, G. (2010). Where did all the GPs go? Increasing supply and geographical equity in England and Scotland. Journal of Health Services Research & Policy, 15(1), 28–35. http://doi.org/10.1258/jhsrp.2009.009003

Gravelle, H., & Sutton, M. (2001). Inequality in the geographical distribution of general practitioners in England and Wales 1974-1995. Journal of Health Services Research & Policy, 6(1), 6–13. http://doi.org/10.1258/1355819011927143

Gruening, G. (2001). Origin and theoretical basis of New Public Management, 4, 1–25.

Harris, K. M. (2003). How do patients choose physicians? Evidence from a national survey of enrollees in employment-related health plans. Health Services Research, 38(2), 711–32. Retrieved from http://www.pubmedcentral.nih.gov/ articlerender.fcgi?artid=1360911&tool=pmcentrez&rendertype=abstract

Harrison, M. I., & Calltorp, J. (2000). The reorientation of market-oriented reforms in Swedish health-care. Health Policy, 50(3), 219–240. Retrieved from http://www.sciencedirect.com/science/article/B6V8X-3YHG866-4/2/1f43bf6fa075cec1477484cd3b6c4bf1

Hart, O., & Moore, J. (1999). Foundations of incomplete contracts. The Review of Economic Studies, 66(1), 115–138. Retrieved from http://restud.oxfordjournals.org/ content/66/1/115.short

Hart, O., Shleifer, A., & Vishny, R. (1997). The proper scope of government: theory and an application to prisons. The Quarterly Journal of …. Retrieved from http://qje.oxfordjournals.org/content/112/4/1127.short

Hefetz, A., Warner, M. E., & Vigoda-Gadot, E. (2014). Concurrent Sourcing in the Public Sector: A Strategy to Manage Contracting Risk. International Public Management Journal, 17(3), 365–386. http://doi.org/10.1080/10967494.2014.935242

50

Hendrikse, G., & Veerman, C. (2001). Marketing Co‐operatives: An Incomplete Contracting Perspective. Journal of Agricultural Economics, 52(1), 53–64. Retrieved from http://onlinelibrary.wiley.com/doi/10.1111/j.1477-9552.2001.tb00909.x/abstract

Hood, C. (1991). A Public Management for All Seasons? Public Administration, 69(1), 3–19. http://doi.org/10.1111/j.1467-9299.1991.tb00779.x

Hood, C. (1995). The “new public management” in the 1980s: Variations on a theme1. Accounting, Organizations and Society, 20(2–3), 93–109. http://doi.org/10.1016/0361-3682(93)E0001-W

Hubbard, A. E., Ahern, J., Fleischer, N. L., Van der Laan, M., Lippman, S. a, Jewell, N., … Satariano, W. a. (2010). To GEE or not to GEE: comparing population average and mixed models for estimating the associations between neighborhood risk factors and health. Epidemiology (Cambridge, Mass.), 21(4), 467–474. http://doi.org/10.1097/EDE.0b013e3181caeb90

Jordahl, H., & Öhrvall, R. (2014). Nationella reformer och lokala initiativ. In H. Jordahl (Ed.), Välfärdstjänster i privat regi: framväxt och drivkrafter (pp. 33–87). Stockholm: SNS förlag.

Kastberg, G. (2008). The Blind Spots of Quasi-market Regulators. Public Organization Review, 8(4), 347–363. http://doi.org/10.1007/s11115-008-0066-3

King, G., Keohane, R. O., & Verba, S. (1994). Designing social inquiry : scientific inference in qualitative research. Princeton N.J.: Princeton University Press. Retrieved from http://www.worldcat.org/title/designing-social-inquiry-scientific-inference-in-qualitative-research/oclc/29225092&referer=brief_results

Koning, P., & Heinrich, C. (2010). Cream-skimming, parking and other intended and unintended effects of performance-based contracting in social welfare services. Retrieved from http://www.econstor.eu/handle/10419/36313

Le Grand, J. (2009). Choice and competition in publicly funded health care. Health Economics, Policy, and Law, 4(Pt 4), 479–88. http://doi.org/10.1017/ S1744133109990077

Le Grand, J., & Bartlett, W. (1993). Quasi-markets and social policy. London: Macmillan. Retrieved from http://libris.kb.se/bib/4831275

Lunander, A., & Andersson, A. (2004). Metoder vid utvärdering av pris och kvalitet i offentlig upphandling - En inventering och analys av utvärderingsmodeller inom offentlig upphandling. Media. Stockholm.

Lundvall, K. (2012). Kvalitetshöjande konkurrens i valfrihetssystem - vad krävs? Stockholm.

Luo, W., & Wang, F. (2003). Measures of spatial accessibility to health care in a GIS environment: Synthesis and a case study in the Chicago region. Environment and Planning B: Planning and Design, 30(6), 865–884. http://doi.org/10.1068/b29120

Marshall, I. J., Wang, Y., Crichton, S., McKevitt, C., Rudd, A. G., & Wolfe, C. D. A. (2015). The effects of socioeconomic status on stroke risk and outcomes. The Lancet Neurology, 14(12), 1206–1218. http://doi.org/10.1016/S1474-4422(15)00200-8

Mendis, D. (2009). Choosing choice? International Journal of Health Planning and Management, 24, 266–275. http://doi.org/10.1002/hpm

Neutens, T. (2015). Accessibility, equity and health care: review and research directions for transport geographers. Journal of Transport Geography, 43, 14–27. http://doi.org/10.1016/j.jtrangeo.2014.12.006

Newhouse, J. P., Williams, A. P., Bennett, B. W., & William, B. (1982). Does the geographical distribution of physicians reflect market failure ? The Bell Journal of Economics, 13(2), 493–505. http://doi.org/10.2307/3003469

51

Oliver, A, & Evans, J. G. (2005). The paradox of promoting choice in a collectivist system. Journal of Medical Ethics, 31(4), 187. http://doi.org/10.1136/ jme.2005.011809

Penchansky, R., & Thomas, W. J. (1981). The Concept of Access: Definition and Relationship to Consumer Satisfaction. Medical Care, 19(2), 127–140.

Pollitt, C., & Dan, S. (2011). The impacts of the New Public Management in Europe: A meta-analysis (COCOPS Working Paper No. No. 3). COCOPS Working Paper. Retrieved from https://lirias.kuleuven.be/handle/123456789/332635

Puig-Junoy, J. (1999). Managing risk selection incentives in health sector reforms. International Journal of Health Planning and Management, 14(4), 287–311. http://doi.org/10.1002/(SICI)1099-1751(199910/12)14:4<287::AID-HPM560>3.0.CO;2-V

Rice, N., & Smith, P. C. (2001). Ethics and geographical equity in health care. Journal of Medical Ethics, 27(4), 256–261. http://doi.org/10.1136/jme.27.4.256

Ringard, A. (2011). Equitable access to elective hospital services: The introduction of patient choice in a decentralised healthcare system. Scandinavian Journal of Public Health. http://doi.org/10.1177/1403494811418277

Robertson, R., & Burge, P. (2011). The impact of patient choice of provider on equity: analysis of a patient survey. Journal of Health Services Research & Policy, 16 Suppl 1(April), 22–8. http://doi.org/10.1258/jhsrp.2010.010084

Rubin, M., Denson, N., Kilpatrick, S., Matthews, K. E., Stehlik, T., & Zyngier, D. (2014). “I Am Working-Class”: Subjective Self-Definition as a Missing Measure of Social Class and Socioeconomic Status in Higher Education Research. Educational Researcher, 43(4), 196–200. http://doi.org/10.3102/0013189X14528373

Sappington, D., & Stiglitz, J. (1987). Privatization, information and incentives. Journal of Policy Analysis and …, 6(4). Retrieved from http://onlinelibrary.wiley.com/doi/10.2307/3323510/abstract

Schillo, S., Lux, G., Wasem, J., & Buchner, F. (2016). High cost pool or high cost groups-How to handle high(est) cost cases in a risk adjustment mechanism? Health Policy (Amsterdam, Netherlands), 120(2), 141–147. http://doi.org/ 10.1016/j.healthpol.2016.01.003

Sclar, E. (2000). You don’t always get what you pay for : the economics of privatization. Ithaca: Cornell University Press.

Segal, I. (1999). Complexity and renegotiation: A foundation for incomplete contracts. The Review of Economic Studies, 66(1), 57–82. Retrieved from http://restud.oxfordjournals.org/content/66/1/57.short

Slyke, D. M. Van. (2013). The for Mythology Social of Privatization in Contracting Services, 63(3), 296–315.

Socialdepartementet. Proposition 2008/09:29 - Lag om valfrihetssystem (2008). Socialstyrelsen. (2010). Stimulansbidrag till valfrihetssystem enligt LOV i

äldreomsorg - Delrapport juni 2010. Stolt, R., Blomqvist, P., & Winblad, U. (2010). Privatization of social services:

Quality differences in Swedish elderly care. Social Science & Medicine (1982), 72(4), 560–567. http://doi.org/10.1016/j.socscimed.2010.11.012

Stolt, R., & Winblad, U. (2009). Mechanisms behind privatization: a case study of private growth in Swedish elderly care. Social Science & Medicine (1982), 68(5), 903–11. http://doi.org/10.1016/j.socscimed.2008.12.011

Tynkkynen, L.-K., Keskimäki, I., & Lehto, J. (2013). Purchaser-provider splits in health care-The case of Finland. Health Policy (Amsterdam, Netherlands), 111(3), 221–5. http://doi.org/10.1016/j.healthpol.2013.05.012

52

von Otter, C., & Saltman, R. B. (1991). Towards a Swedish health policy for the 1990s: planned markets and public firms. Social Science & Medicine (1982), 32(4), 473–81. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/2024163

Walsh, K., Deakin, N., Smith, P., Spurgeon, P., & Thomas, N. (1997). Contracting for change : contracts in health, social care, and other local government services. Oxford: Oxford University Press. Retrieved from http://libris.kb.se/bib/4627356

Williamson, O. (1975). Markets and hierarchies, analysis and antitrust implications : a study in the economics of internal organization. New York: Free Press.

Winblad, U., Andersson, C., & Isaksson, D. (2009). Kundval i hemtjänsten - Erfarenheter av information och uppföljning. Stockholm.

Winblad, U., Isaksson, D., & Bergman, P. (2012). Effekter av valfrihet inom hälso- och sjukvård – en kartläggning av kunskapsläget. Stockholm: Myndigheten för vårdanalys.

Acta Universitatis UpsaliensisDigital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Medicine 1473

Editor: The Dean of the Faculty of Medicine

A doctoral dissertation from the Faculty of Medicine, UppsalaUniversity, is usually a summary of a number of papers. A fewcopies of the complete dissertation are kept at major Swedishresearch libraries, while the summary alone is distributedinternationally through the series Digital ComprehensiveSummaries of Uppsala Dissertations from the Faculty ofMedicine. (Prior to January, 2005, the series was publishedunder the title “Comprehensive Summaries of UppsalaDissertations from the Faculty of Medicine”.)

Distribution: publications.uu.seurn:nbn:se:uu:diva-354476

ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2018