2.1 Changes in the role of primary care

Changes in the overall role of primary care across the five county councils following reform are summarised in Table 1. Focusing on the three county councils that implemented their new models in 2007 and 2008, important differences exist between Halland on the one hand and Västmanland and Stockholm on the other. Halland, which was the first county council to introduce reform in primary care, had a more progressive vision to change the overall health-care structure in favour of primary care. The objective was that 80% of all outpatient visits should be taken care of by primary care providers. This was a rather ambitious objective by Swedish standards. The gate-keeping role of primary care has been limited and hospitals are still responsible for a significant share of outpatient visits to doctors (Glenngård et al., Reference Glenngård, Hjalte, Svensson, Anell and Bankauskaite2005). In 2006, before the introduction of reform, primary care providers in Halland were responsible for about 67% of all outpatient care visits on average, which was somewhat higher than the national average. The objective to expand the role of primary care had a direct influence on the payment system. Fixed payment to providers for each registered individual in Halland has been reduced from the start of reform if the degree of coverage (so called ‘täckningsgrad’) is below 80%. Primary care providers in Halland are also encouraged to contract or employ specialist doctors besides GPs as well as specialist nurses to fulfil this objective. Formal regulation of how primary care services should be organised or what type of staff providers should employ or contract has been kept at a minimum (Landstinget Halland, 2008).

Table 1

Comparison between primary care reforms in five county council’s

The organisation of primary care in Stockholm has been more regulated. Employment of doctors other than GPs and paediatric and geriatric specialists are prohibited (Stockholms läns landsting, 2008). In both Västmanland and Stockholm, the objective of reforms implemented in 2008 was to improve access to primary care services within its existing role, rather than to introduce changes in the overall health-care structure.

An important difference compared to both Halland and Västmanland is that Stockholm distinguishes between five different primary care services: the family doctor services, child health care, maternity care, care of the feet and speech therapists. By developing separate accreditation and payment systems for these services, it was argued that more private alternatives and increased possibilities for the population to choose were created. In contrast, both Västmanland and Halland gave providers a broad responsibility to deliver primary care services, arguing that this was beneficial especially for chronic patients in need of multidisciplinary services.

Implementation of reform in Region Skåne and the VG-region in 2009 in principle followed the model in Halland. The main difference was a lower initial ambition regarding the degree of coverage. In Region Skåne, for example, the standard payment assumes that 55–61% of all outpatient care visits are taken care of by primary care providers. If this is not the case, the payment is adjusted accordingly.

2.2 Choice for the population

In all five county councils, an important part of reform was to facilitate choice and competition by allowing more private providers to enter the primary care market. Västmanland relied on a public tendering process whereby accepted private providers were allowed to start up new practices during 2008–2011. In the other four county councils, new private providers have been accredited on a continual basis if they fulfil and accept qualification requirements and responsibilities.

In all five county councils, funding is not guaranteed but depends on how well providers compete for individual registrations. In Halland, individuals have to be registered with a primary care unit, either by active choice or by acceptance of an alternative suggested by the county council. When the new model was implemented, inhabitants received a letter from the county council proposing registration with a primary care provider based on either previous visits or geographical distance (in that order). If individuals did not oppose this suggestion, they were registered accordingly. The main argument in support of registration with a unit rather than a GP was that the county council should regulate as little as possible on how providers organise contacts with their patients. Similar principles are used in Västmanland, Region Skåne och the VG-region. In Stockholm, registration is with GPs rather than units and there is no passive registration of patients who do not make an active choice.

In none of the five county councils are there any formal limits regarding the maximum or minimum number of registered individuals for each unit or GP. The possibilities for providers to deny individuals a place on their list is limited. In principle, each provider should expand their services if more people choose to register with them and all new registrations must follow a first-come, first-service basis. There are no formal limits regarding the number of times individuals can choose to delist and register with an alternative provider.

Although individual patients are encouraged to use services where they have registered, patients are in fact allowed to seek care at other (i.e. competing) primary care providers or from private specialist practices, if available. This matches traditions in Swedish health care, with possibilities for the population to seek care from a broad range of health-care services directly, and a limited formal gate-keeping role for primary care (Kroneman et al., Reference Kroneman, Maarse and van der Zee2006). In both Region Skåne and the VG-region, user charges are differentiated to give patients financial incentives to be ‘loyal’ to the provider where they have registered. In Region Skåne, for example, patients pay 120 SEK if they visit a doctor at the unit where they have registered and 200 SEK for a similar visit at other primary care units. In the other three county councils, user charges for all primary care services are the same and about 150 SEK for a doctor visit. In all five county councils, a higher user charge applies for visits to specialists without a referral, about 300 SEK. Owing to a 900 SEK high-cost ceiling for visits during a 12-month period, several patients in practice have a ‘free card’, which means that they are not affected by differentiated user charges.

2.3 Payment system and financial responsibility

Differences between the county councils in terms of payment system and financial responsibility for providers can in part be explained with reference to objectives behind the reform. In the four county councils where providers are responsible for a comprehensive primary care service, that is, Halland, Västmanland, Region Skåne and the VG-region, providers have a broad financial responsibility for all expenditures related to the delivery of primary care. At least 80% of the provider’s income is based on fixed annual payments per registered individuals (see Table 1). In principle, providers in these county councils have a fundholding responsibility for primary care expenditures related to their subscribers, irrespective of where the consumption of services takes place. This means that a provider may have to pay other providers if patients choose not to be loyal to where they are registered. If providers do not agree voluntarily on a transfer price, a payment schedule determined by the county council applies. In both Region Skåne and the VG-region, the fundholding responsibility also includes expenditures related to general prescription drugs. By comparison, primary care providers in Stockholm have a more limited financial responsibility and receive about 60% of their income based on pay per visits to GPs and other staff.

Key arguments against payment per visit is that providers then have financial incentives to increase the volume of visits rather than the quality or value of care, and to shorten visits and give priority to easy cases and repeat visits rather than to allocate resources according to the health-care needs of patients (Robinson, Reference Robinson2001). This argument was also important in the four county councils that rely more extensively on a fixed payment per registered individual. It was also argued that payment per visits preserved a traditional organisation of practices and did not stimulate new and innovative ideas on how to organise contacts with patients. In both Region Skåne and the VG-region, this argument was taken to its extreme as providers do not get any payment based on the volume of visits.

Several important differences also exist between the four county councils in respect of fixed and comprehensive payment (see Table 2). In Halland and Västmanland (as in Stockholm), more simple models for risk adjustment are used based on age groups. Indeed, this has created tensions and debate regarding ‘unfair’ payments, particularly in Stockholm where payment to providers previously was adjusted based on socio-economic indicators for the geographical area where providers were located. In both Region Skåne and the VG-region, more advanced models for risk adjustment based on classification of diagnosis using the ACG (adjusted clinical groups) system (Majeed et al., Reference Majeed, Bindman and Weiner2001) in combination with socio-economic indicators are applied.

Table 2

Comparison of payment system for primary care year 2009 in five county council’s

A small proportion of payment in all five county councils is based on performance. In Halland, each provider who did not participate in a national diabetes quality register had to pay a lump-sum penalty in 2008. For 2009, a similar penalty payment exists for providers with poor compliance towards five prescribing targets (e.g. use of generic drugs and antibiotics) developed by the county council drug formulary committee. In Stockholm, pay-for-performance is organised in several sub-systems. A lump-sum penalty payment exists for practices that do not have an environmental certificate or do not provide the county council with information agreed upon. About 3% of the total annual payment is linked to targets such as access to practices over the phone, results from patient satisfaction surveys, registration in the national diabetes quality register and compliance towards recommendations from the county council drug formulary committee. Similar principles of pay-for-performance using 7–12 targets related to access over the phone, recommendations from the drug formulary committee, preventive services, registration in the national diabetes quality register, clinical process indicators (diabetes and asthma) and/or patient satisfaction exist in Västmanland, Region Skåne and the VG-region.

4.1 ‘Cream-skimming’ and under-provision of services

Problems with fixed annual payments for each registered individual include the need to risk-adjust payments to prevent a situation where providers ‘cream-skim’ to avoid the sick and unhealthy (see e.g. Robinson, Reference Robinson2001; Gosden et al., Reference Gosden2004). Even if providers are not allowed to deny individuals a place on their lists, which is the case in all five county councils, providers may try to influence who will register indirectly, for example, by selecting a preferable geographical location and/or by discriminating marketing activities. Such a ‘cream-skimming’ behaviour can be avoided if county councils risk-adjust capitation payments. It can be concluded that county councils differ with respect to ambitions in this area. So far, however, there is no knowledge about the extent to which more advanced models for risk adjustment in Region Skåne and the VG-region will have an impact on provider behaviour and the distribution of services across patient groups.

Even with a risk-adjusted fixed payment, there is still a risk that providers will skimp on services and under-provide care since all things done to patients will increase expenditures and reduce the provider’s profits. Studies in both Denmark and Norway also show that more reliance on capitation payment is associated with an increased number of referrals to specialists (Krasnik et al., Reference Krasnik, Groenewegen, Pedersen, Scholten, Mooney, Gottschau, Flierman and Damsgaard1990; Iversen and Lurås Reference Iversen and Lurås2000). With a fixed payment, providers benefit financially if patients (and the associated health-care expenditure) can be referred elsewhere. With more payment based on the volume of services, GP practices are instead given financial incentives to take care of their patients themselves.

A cost-shifting behaviour can be mitigated by enforcing a financial responsibility if patients are referred to other providers. At least in part, such a fundholding responsibility for primary care providers is also being used by Swedish county councils. With a broad financial responsibility, however, providers are incentivised not only to under-provide their own services, but also to limit access to health-care services more generally. Possibilities to limit access exist more directly for diagnostic services that require a referral from a GP. In those county councils that practise a financial responsibility for prescriptions, primary care units may also under-provide medicines.

Owing to financial incentives to limit access and skimp on services, a comprehensive and fixed payment may also have a negative effect on the trust in primary care on the part of the population. A previous comparison across OECD countries suggests that satisfaction with GP services is lower in countries with a formal gate-keeping role of primary care and with more limited direct access to specialist services for the population (Kroneman et al., Reference Kroneman, Maarse and van der Zee2006). A possible explanation can be deduced using agency theory. GPs in general have previously been defined as ‘double agents’ who should consider both expectations from individual patients as well as the need for priority setting from a societal perspective (Blomqvist, Reference Blomqvist1991). On the one hand, GPs should provide patients with services of good quality and easy access to relevant specialists. On the other, they have an organisational role to prioritise and avoid unnecessary referrals to specialist care. For GPs with a comprehensive and ‘tight’ financial responsibility, it will be more difficult to match expectations from patients and a gap in trust between primary care and the population is more likely to develop.

There are several possible ways to balance the financial incentives that follow from comprehensive fixed payment, for example, enforcement of clinical guidelines, exceptions for patients with high medical costs (outliers), transparent comparison of results across providers and pay-for-performance based on quality indicators (Miller, Reference Miller2009). In theory, competition based on choice should also counterbalance financial incentives to under-provide services. If choices are real and informed, primary care providers who hold back on services can be avoided by the population. This, in turn, gives financial incentives for units to provide more balanced services. As providers in the Swedish primary care market are unable to influence price (payment from county councils and user charges for patients are fixed), profit levels will ideally be reduced to a reasonable level as providers compete on improvements in the quality and quantity of care. For county councils that believe in this logic and the benefits of competition, it will first of all be important to encourage and facilitate choice for the population. The second and equally important condition is that the population have access to alternative primary care providers to choose from (Le Grand, Reference Le Grand2007).

4.2 Comprehensive fixed payment and conditions for competition

In traditional economic textbooks, the theoretical model of perfect competition is often associated with many buyers and sellers and unrestricted access to information about the price and quality of products or services. In practice, both theoretical and empirical studies show that a competitive situation that is similar to the model of perfect competition in terms of outcomes can be achieved with 5–6 or even a lower number of alternative providers (Hultkrantz and Nilsson, Reference Hultkrantz and Nilsson2004). If alternatives differ only by price, studies suggest that a price equal to the price that would have been established with perfect competition will be set even with only two alternative providers, the so-called Bertrand competition. With more differentiated products, more providers are needed. More than two providers are also necessary to counteract the development of silent cartels where providers collaborate without engaging in direct negotiations.

Perhaps more important than the number of alternative producers on any given market (a static perspective) is the possibility for new competitors to enter the same market (a dynamic perspective) (Shapiro and Varian, Reference Shapiro and Varian1999). New competitors often bring innovative ideas to the market and support a dynamic development. A threat from new potential competitors also puts additional pressures on existing alternatives, even if they are few in number, the so-called contestability (Baumol et al., Reference Baumol, Panzar and Willig1982).

As county councils will have a key role as single payers in the local primary care markets in Sweden, local policies to preserve and strengthen competition will indeed be important. Basic measures to facilitate competition will be the same irrespective of which primary care model county councils implement. Choice for the population needs to be encouraged, supported and informed, and barriers for new providers to enter the primary care market should be kept at a reasonable low level.

To encourage, support and inform the population is likely to be a similar challenge for all county councils. Evidence suggests that individuals in general do want the option to choose their primary care provider (Scott, Reference Scott2000; Hjelmgren and Anell, Reference Hjelmgren and Anell2007; Le Grand, Reference Le Grand2007) but are more sluggish when it comes to shifting between providers in practice (Mold et al., Reference Mold, Fryer and Roberts2004; Fotaki et al., Reference Fotaki, Roland, Boyd, McDonald, Scheaff and Smith2008). When it comes to supporting alternative providers, however, conditions for competition will at least to some extent be influenced by characteristics of the model implemented.

It can be argued that the principle of a fixed and comprehensive payment used by a majority of the Swedish county councils will increase barriers for new providers to enter the primary care market. A fixed payment means that each provider will have to bear a higher financial risk, since an increase in the quantity of care is not linked to any increase in revenue (Jegers et al., Reference Jegers, Kesteloot, De Graeve and Gilles2002). With pay-per-visit, and even more so with fee-for-service, the financial risk will instead be on the county councils. Furthermore, with a comprehensive financial responsibility for providers, variation in expected primary care expenditures across patients will increase compared to a situation with more limited financial responsibility. A risk-adjusted payment will indeed give more resources to providers who register more patients with chronic conditions. But variation in actual expenditures within each risk-adjusted group of patients will still exist. If providers have responsibility for a full range of primary care services including expenditures for prescription drugs, this variation will be more significant.

The easiest way for primary care providers to manage non-systematic variation in health-care expenditures is to grow in size by registration of more individuals or to integrate horizontally with other providers. With a larger total list of registered individuals, the ‘law of big numbers’ means that differences between actual expenditures and payments at the individual level will balance each other out. Risk-adjusted payment for patients with a higher consumption of services than expected (given their risk level) will most likely be subsidised by registered individuals with a lower consumption than expected. This also poses a dilemma in terms of governance. On the one hand, it will be more important for county councils that have chosen a fixed and comprehensive payment to facilitate competition across providers. This is to counterbalance the potential problems of skimping and under-provision. On the other hand, it may be more difficult to develop such a market where patients can delist from providers with poor performance, since the payment system and financial responsibility favour larger units and/or horizontal integration and thereby market concentration. Smaller units will have a disadvantage because of the financial risk involved. In particular, it may be difficult for new providers to enter the market, since patients who first register are more likely to be dissatisfied with their previous provider and more demanding than the average. Ironically, the model originally developed in the rural county council of Halland probably fits the general competitive conditions in the more densely populated Stockholm region best. The model developed and implemented in Stockholm, with more pay per visit and limited financial responsibility for providers, has competitive advantages in rural areas with more limited room for alternative providers.