1. Introduction
Chronic diseases are major causes of death and reduction in the quality of life worldwide, and their prevalence is expected to rise. Disease management programmes (DMPs) have been presented as a policy response to challenges of care coordination for such chronic diseases. Although the concepts and designs of DMPs differ (Nolte and Hinrichs, Reference Nolte and Hinrichs2012), they share a common aspiration of developing and formalizing processes to support patient journeys and facilitate collaboration in the management of chronic conditions (Tsiachristas et al., Reference Tsiachristas, Cramm, Nieboer and Rutten-van Mölken2013a, Reference Tsiachristas, Dikkers, Melinde, Boland and Rutten-van Mölken2013b; Schang et al., Reference Schang, Thomson and Czypionka2016). DMPs may, therefore, be seen as an illustration of a broader trend in health policy and management aiming to structure the complexity of interactions through administrative procedures, standards and detailed guidelines for health care professionals and organizations (Hogle, Reference Hogle1995; Timmermans and Berg, Reference Timmermans, Berg, Timmermans and Berg1998; Brunsson, Reference Brunsson, Egeberg and Lægreid1999; Busch, Reference Busch2011). Complexity in health care can be perceived as a situation where there is high number of interacting components that influence the delivery and outcome of services. Other defining characteristics include the (high) number and difficulty of behaviours required by those delivering or receiving the intervention, the involvement of several organizational levels, variability of outcomes and flexibility or tailoring of the intervention (Craig et al., Reference Craig, Dieppe, Macintyre, Michie, Nazareth and Petticrew2008).
DMPs aim to address such complexity through detailed descriptions of the sequence and responsibilities for diagnostic, treatment and rehabilitation procedures as well as (secondary or tertiary) prevention activities. This is supposed to reduce duplication in services and to improve efficiency, patient experiences and health outcomes. Although the broad aspirations are similar across countries, it is clear that the actual results will depend on the translation and implementation of the programme into regulatory instruments and practices in specific health care settings (Christensen et al., Reference Christensen, Lægreid, Roness and Røvik2007; Schang et al., Reference Schang, Thomson and Czypionka2016).
This paper investigates the implementation of DMPs in a National Health Care system in the Nordic region using type II diabetes as an example. Denmark has taken a broader and more systemic approach to DMPs compared to many other countries emphasizing the involvement of health care actors at different levels and adopting a process of gradual and negotiated implementation allowing for interpretations and development of the concept over time. Although this approach has advantages, it also carries an inherent risk of losing momentum and diffusing the perspective into pragmatic application of what is acceptable for the different implementation stakeholders. To explore implementation processes in the Danish case, we take a long-term systemic perspective covering the period from 2008 to 2020.
The aim is to extend the scientific knowledge about DMP perspectives and implementation beyond the previous case studies of Germany and Austria (Ostermann et al., Reference Ostermann, Hoess and Mueller2012; Schang et al., Reference Schang, Thomson and Czypionka2016) and international reviews (Nolte et al., Reference Nolte, Knai and McKee2008, Nolte and Hinrichs, Reference Nolte and Hinrichs2012). To achieve this, we apply an analytical framework for DMP implementation that takes an integrated, systemic perspective (Winter and Nielsen, Reference Winter and Nielsen2008) emphasizing the role of key stakeholders (Schang et al., Reference Schang, Thomson and Czypionka2016). Studying DMP implementation can provide input to understanding the diversity of positive (Weingarten et al., Reference Weingarten, Henning, Badamgarav, Knight, Hasselblad, Gano and Ofman2002; Ofman et al., Reference Ofman, Badamgarav, Henning, Knight, Gano, Levan, Gur-Arie, Richards, Hasselblad and Weingarten2004; Nolte and Hinrichs, Reference Nolte and Hinrichs2012) and mixed effects of the significant investments in DMPs (Cretin et al., Reference Cretin, Shortell and Keeler2004; Campbell et al., Reference Campbell, Murray, Darbyshire, Emery, Farmer, Griffiths, Guthrie, Lester, Wilson and Kinmonth2007; Lemmens et al., Reference Lemmens, Nieboer, van Schayck, Asin and Huijsman2008). It can also contribute to broader debates about implementation challenges and strategies for complex programme interventions in health care systems.
2. Theory and methods
Many implementation studies take a ‘top-down’ perspective (Pressman and Wildawsky, Reference Pressman and Wildawsky1973; Sabatier and Mazmanian, Reference Sabatier and Mazmanian1979; Sabatier, Reference Sabatier1986; Schang et al., Reference Schang, Thomson and Czypionka2016). Others prefer a ‘bottom-up’ perspective emphasizing the discretionary power of ‘street level bureaucrats’ (e.g. Hjern and Hull, Reference Hjern and Hull1982). Recent contributions argue for an integrated approach that looks at the interaction of stakeholders within institutional contexts in shaping critical implementation dimensions of policy design, organizational and interorganizational behaviour, management, and responses from street-level-bureaucrats and target groups (Winter and Nielsen, Reference Winter and Nielsen2008). Based on this approach, implementation can be seen as a string of interpretation processes and decision points, which determine policy practices. Decision points produce decision artefacts such as action plans, budgets, agreements, guidelines, payment schemes, etc. Stakeholders influence decision-making and subsequent interpretation of decision artefacts to form the actual implementation patterns (Czarniawska and Sevon, Reference Czarniawska and Sevon1996; Christensen et al., Reference Christensen, Lægreid, Roness and Røvik2007; Lemmens et al., Reference Lemmens, Nieboer, van Schayck, Asin and Huijsman2008; Tsiachristas et al., Reference Tsiachristas, Cramm, Nieboer and Rutten-van Mölken2013a, Reference Tsiachristas, Dikkers, Melinde, Boland and Rutten-van Mölken2013b). This means that policy intentions are renegotiated and reshaped in a continuous and interactive process during the implementation phase (Juul Nielsen and Langstrup, Reference Juul Nielsen, Langstrup, Mayer, Carpes and Knoblich2014; Richards, Reference Richards, Richards and Hallberg2015) and that deviations and recursivity can be expected.
The study relies on desk top analysis of national and regional policy documents and secondary data sources, combined with new survey material from Danish municipalities.
We use content analysis to trace implementation processes according to dimensions derived from the ‘programme theory’ of the main DMP document (see the next section). In contrast to DMP models in other countries, this document presents a rather broad formulation of systemic direction, rather than a narrow assignment of responsibility e.g. to general practitioners (GPs). This means, that analysis of DMP implementation must account for actions of several stakeholders at different governance levels. Our analysis section is, therefore, structured in two parts, where the first part presents documentation about stakeholders, and the second part summarizes across stakeholders using the analytical dimensions derived from the ‘programme theory’.
As stated in the Introduction, we perceive implementation as a string of interpretation processes and decision points, which determine policy practices. Decision processes produce decision artefacts such as action plans, budgets, agreements, guidelines, payment schemes, etc. We, therefore, base the analysis on collection of official documents that present such decision artefacts referring to the DMPs and associated principles.
The national and regional DMP documents were summarized and compared in Utoft and Kristensen (Reference Utoft and Kristensen2019). Although these documents present a useful overview of the intentions at the national and regional levels, it is clear that the actual implementation depends on subsequent interpretations and reactions by hospitals, GPs and municipalities. Several methodological challenges emerge, when moving to these levels. Most importantly, it appears that many initiatives are integrated into other ongoing policies and agreements aiming to shape services for chronic care patients and that the awareness of the link to specific national and regional DMP formulations can be weak among actors ‘on the ground’. Our approach to solving these issues is to track the four DMP implementation dimensions using several other data sources including (a) policy documents dealing with specific implementation issues such as regional-municipal agreements, and national agreements for GPs, (b) secondary studies of implementation relevant issues and (c) a survey administered to the municipalities.
The municipality survey was designed for this purpose and refers to DMPs and DMP-related activities. It was distributed to bureaucrats in all municipalities in 2019 and obtained a response rate of 88%.
Ideally, we would have conducted a similar survey among GPs. However, exploratory interviews with selected GPs revealed a high degree of confusion about the relationship between the many ongoing initiatives related to chronic care optimization and the formulation of the national and regional level DMP programmes. In combination with the notoriously low propensity to respond to general surveys among GPs, this led us to opt for other data sources for this group of implementation stakeholders.
This ‘patchwork’ of many data sources is useful for uncovering the implementation of DMPs in Denmark. Yet, there are also limitations. Most importantly, the ambiguities in the pathway from policy formulation to implementation, and the many concurrent initiatives in the period mean that caution should be taken when assessing the success of the programme. We present evidence of the implementation process and provide indications of the general progress in diabetes care in the period from 2008 to 2020. We cautiously suggest that the progress can at least partly be related to the introduction of the structured DMP approaches.
3. Policy design, ‘programme theory’ and analytical dimensions for the implementation process
The main document for DMPs in Denmark is the generic DMP model, which was published by The Danish Health Authority in 2008. The Danish Health Authority is a national agency under the Ministry of Health, which is responsible for general planning, guidelines and standards for health care, and surveillance and licensing of health care personnel. The DMP description refers to the ‘Chronic Care Model’ as a key inspiration and elements of the Chronic Care Management (CCM) are translated into ideas about stratification and formulation of standard treatment pathways.
The generic DMP model in Denmark defines DMPs as ‘a description of the totality of multiprofessional, cross-sectorial and coordinated health care interventions for a given chronic condition, which secures the application of evidence based recommendations for health professional activities and a precise description of the division of labor and the communication and coordination between all involved actors’ (Danish Health Authority, 2008, own translation).
The specific components of the model include: determination of the target population; specification of interventions with reference to evidence-based clinical guidelines; determination of the organization, division of labour and stratification; establishment of standards and indicators to monitor quality and effects and a plan for implementation, evaluation and follow-up.
The individual components can be seen as mechanisms through which the DMP is supposed to achieve its aim of providing ‘…improved patient satisfaction, better patient compliance and disease control and more efficient use of resources’ (Danish Health Authority, 2008: 12, own translation). However, it is also acknowledged that DMPs may be implemented in different ways (Danish Health Authority, 2008: 12). Indeed, the document is presented as a generic template, while the specific elements should be designed for different chronic diseases and according to conditions at the regional and local levels. Overall, the generic model and associated policy documents present an implicit ‘programme theory’ (Leeuw, Reference Leeuw2003) and sets the parameters for subsequent interpretation and implementation. Using the process dimensions from the DMP ‘programme theory’ (as described above and summarized into the four analytical concepts of stratification, changing practices, governance infrastructure and alignment of incentives and norms) and a systemic stakeholder perspective (Winter and Nielsen, Reference Winter and Nielsen2008) enables a comprehensive systemic view of the implementation process.
The following analysis, therefore, consists of two parts. First, a stakeholder analysis to illustrate the systemic perspective of the implementation efforts, and second, a thematic analysis based on the four key implementation dimensions presented in Table 1.
Process dimensions for analysis of implementation of DMPs
4. Stakeholders and implementation behaviour
The Danish health system can be characterized as universal, mostly public and decentralized. Management and delivery of health services is the responsibility of five Danish regions and 98 municipalities based on regulatory frameworks from the state (Olejaz et al., Reference Olejaz, Nielsen, Rudkjøbing, Birk, Krasnik and Hernández-Quevedo2012). This means that regions, hospitals, GPs and municipalities are key stakeholders for implementation of DMPs.
4.1 Regions
The five regions manage regional hospital systems and are in charge of planning and payment of primary care provided by private GPs. The regions are mainly funded by block grants allocated from the state according to socio-economic criteria and annual budget negotiations.
All five regions developed DMP implementation plans between 2008 and 2012, and have subsequently published revised versions in response to a new generic template from 2012 (Utoft and Kristensen, Reference Utoft and Kristensen2019).
In spite of the common template, there is some variation in the specific design of the regional DMPs (Utoft and Kristensen, Reference Utoft and Kristensen2019). These differences mainly relate to the stratification of patients where two regions include own care ability in addition to the standard assessment of disease severity, complications and multimorbidity. Three out of the five regions indicate that they undertake dynamic and ongoing stratification assessments.
In accordance with the national template, all five regions assign key roles to the GPs that are given the main responsibility for early detection and diagnosis. GPs are meant to do the initial stratification of patients and to refer to other specialists when relevant. For most patients, GPs should remain the primary health care professional for disease management and life style counselling, which is also done in the municipalities as part of their obligation to provide rehabilitation and prevention services. Physical activity, smoking, diet and alcohol is explicitly mentioned for three regions, while the two others use more general terms with regards to life style interventions.
Two features of regional governance are particularly important for the implementation of DMPs. First, the non-hierarchical relationship between regions, municipalities and private GPs means that the primary instruments for alignment are negotiated national agreements between regions and GPs and mandatory ‘health agreements’ between regions and municipalities. The health agreements include collaboration and care integration issues as outlined in the DMPs. Guidelines and standards for collaborative practices are also included as part of the agreements.
Second, from 2007 to 2019 there was a strong national policy emphasis on regional productivity with a mandatory 2% activity increase per year. This implied an incentive for the regions to maintain activity within their own hospitals rather than pursuing solutions that would ‘export’ patients to GPs (Højgaard et al., Reference Højgaard, Kjellberg and Bech2018). The productivity focus also meant that activities that could support systemic DMP activities such as phone consultancy between specialists and GPs, follow-up upon discharge or telemedicine have not been incentivized within the diagnosis related groups (DRG) system.
In recognition of the negative side effects, the government abandoned the 2% annual productivity increase requirement in 2019. At the same time, a new ‘proximity funding pool’ was introduced as a replacement for the previous national ‘activity funding pool’. The ‘proximity funding pool’ (1.5 billion DKK) is paid to the regions based on five criteria: reduction in the number of hospital contacts per citizen, reduction in hospital activity for patients with diabetes and/or chronic obstructive pulmonary disease (COPD), reduction in acute readmissions and increased use of telemedicine. The regions are supposed to collaborate with the municipalities and GPs to make progress towards these goals. The intention of this change is aligned with the DMP principles of transferring chronic care activities to GPs (and municipalities) and improves management and coordination through digital solutions and telemedicine.
4.2 Hospitals
The vast majority of Danish hospitals are owned by the regions and operate according to management practices within each of the five regions. Until 2019, the regions generally pushed the national requirement of 2% annual productivity increases on to the hospitals creating a strong impetus to focus on activity targets (Højgaard et al., Reference Højgaard, Kjellberg and Bech2018). Some regions have also incentivized process quality measures (waiting times, etc.), but usually not with specific incentives regarding coordination with GPs and municipalities (McKinsey and Company, 2017). Consequently, some of the aspirations of using DMPs to improve coordination and move activities out of hospitals have not been incentivized within the regions (McKinsey and Company, 2017). Nonetheless, many hospitals have initiated outgoing activities such as hotlines for patients, GPs and municipalities, educational activities, home-based follow-up upon discharge, telecare solutions, etc. Such activities have typically been fostered by regional/municipal agreements and project-based financing rather than system level incentives. Professional networks, personal engagement and ‘grass root’ activities have been important drivers at the project level (McKinsey and Company, 2017).
Similar to GPs, the hospitals are subject to national clinical guidelines and standards. Clinical guidelines for diabetes treatment in hospitals have been issued by the Danish Endocrinological Society in 2011 and 2014 (general treatment), 2012 (surgery), 2015 (neuropathy) and 2014 (lower extremities). Adherence to the guidelines was central in the comprehensive, accreditation-based ‘Danish Quality Management Program’ (DQMP) from 2004 to 2018. The accreditation criteria also included collaboration and communication with external partners during admission and upon discharge and can be seen as part of the effort to support a structured approach to interaction. The DQMP has subsequently been replaced by a more flexible system of national and regional/local quality targets within eight overall themes (chronic care, integrated care, patient safety and survival, quality, process, patient involvement, life expectancy and efficiency) after critique from health professionals and hospitals (Ministry of Health, Danish Regions and Local Government Denmark, 2018). DMPs and health agreements are mentioned as supporting structures in the pursuit of the new targets (Ministry of Health, Danish Regions and Local Government Denmark, 2018). This illustrates how the idea of DMP as a general reference point is maintained at the decision apex, while the policy idea is connected to different specific initiatives over time.
4.3 General practice
Danish GPs operate within private solo or multi practices receiving most of their income through public funding. They are paid on a mixed capitation (30%) and fee for service (70%) basis. Fees are negotiated in national agreements between the association of GPs and Danish Regions (the association representing the five regions). The collective agreement means that GPs have a clear incentive to increase activity, including diabetes-related activity until a predefined collective threshold is reached.
A pilot project with fixed annual payment for diabetes management was initiated in 2007 to promote a more holistic approach to chronic care. The pilot was based on voluntary enrollment, and at the end of the period (2007–2012) approximately 30% of practices had enrolled extending services to approximately 10% of the diabetes population (Rudkjøbing et al., Reference Rudkjøbing, Vrangbaek, Birk, Andersen and Krasnik2015). The pilot project was abandoned in 2012 and a conflict between GPs and the regions restrained follow-up initiatives between 2012 and 2017. A new agreement for 2018 reintroduced a mandatory (but gradual) transition to a new ‘chronic care management fee’ for GPs. According to this scheme, GPs receive a lump sum annual fee to cover all contacts for patients with diabetes, COPD or diabetes + COPD. The fee was determined as an average of the fees paid for GP services to these patient groups in the year prior to the 2018 agreement plus 22% to account for the expected increase in the number (and complexity) of diabetes patients as they are moved from hospital to GP management. The fee is paid on a monthly basis, and if a patient is referred to hospital due to acute deterioration in condition, then payments are suspended until the patient is stabilized and responsibility shifts back to the GP.
GPs are required to follow clinical guidelines and to develop individual patient management plans (PMPs) for diabetes (type II) and COPD patients as part of the scheme. PMPs are digital plans that are established, used and maintained by GPs with the aim of providing a better platform for patient involvement and case management.
The data collected along with the PMPs is reported to national level clinical databases, and is to be used for data-supported learning, which is also facilitated through mandatory participation in ‘quality clusters’ with other GPs (Overenskomst om almen praksis, 2018).
The national GP agreement from 2018 repeats the intention of gradually moving the main responsibility for approximately 25,000 additional patients with diabetes from hospital ambulatory clinics to GPs (10% of all T2DB patients). This is to be implemented according to regional plans, and is accompanied by an upscaling of GP access to specialist support in hospitals (Danish Medical Association, 2019).
From 2018 to 2019, the proportion of citizens with diabetes and diabetes + COPD that were registered within the GP chronic care fee scheme increased from around 40% to around 60% (Sundhedsdatastyrelsen, 2020). This means that diabetes patients are gradually being migrated into the scheme, although with considerable differences across GP practices (Sundhedsdatastyrelsen, 2020).
Citizens included in the fee scheme had fewer contacts with GPs compared to those outside the scheme in 2018, and also slightly fewer contacts with hospitals. Most of the hospital contacts were planned, ambulatory visits (Sundhedsdatastyrelsen, 2020). It appears that the scheme contributes to reversing the trend that was seen in previous years, where both GP and hospital activity increased for chronic care patients from 2010 to 2014 in spite of the aspirations to shift the locus of chronic care from hospitals to GPs and municipalities (McKinsey and Company, 2017).
Another pillar for DMP implementation in GP practice is clinical guidelines for diabetes patients. Such guidelines have been issued by the Danish Health Authority and the Medical Society for General Practice and the Danish Society for Endocrinology in 2011, 2012 and 2014 [Danish Society for General Practice (DSAM), 2012; Ministry of Health, 2016]. These guidelines were part of the accreditation process, which most GPs underwent between 2016 and 2018. The accreditation process also evaluated GPs' knowledge of DMPs and guidelines for referrals, etc. (IKAS 2016). Yet, these data were not reported beyond the accreditation process, and there is no publicly available information about the actual adherence to DMPs or clinical guidelines.
Another part of the implementation strategy is the expectation that GPs comply with the political ‘health agreements’ between regions and municipalities, which provide frameworks for coordination of services. Yet, their incentives to do so are limited, as they are not a formal partner and there are no steering mechanisms or clear incentives for the GPs in the agreements (McKinsey and Company, 2017).
Some additional indications of GP behaviour can be found in studies published in the period. First, it has been documented that the level of referrals from GPs to municipal rehabilitation and care services has been lower than expected. A study from 2014 showed that 70% of the GPs had referred five times or less to municipal rehabilitation services for diabetes and 13% had not referred to municipalities at all (Andersen et al., Reference Andersen, Thorsen and Kousgaard2014). When asking the non- or low-referring GPs about the reasons they pointed to limited knowledge of municipal services (28%) and uncertainty about the benefits for their patients (24%) (Andersen et al., Reference Andersen, Thorsen and Kousgaard2014). Others pointed to the lack of motivation from patients or explained that GPs were better positioned to provide counselling about life style. Overall, this shows a diversity of GP approaches and variability in knowledge and/or trust in the quality of municipal services in spite of the aspirations to promote collaboration within the DMP framework. Qualitative data collected from GPs confirm the picture of variability in GP knowledge of municipal offers and diversity in motivation (Andersen et al., Reference Andersen, Thorsen and Kousgaard2014). The GPs also expressed frustration with e-referral procedures and the level of information from municipalities after patients complete the programmes. This is surprising given that all municipalities claim to provide this information (Andersen et al., Reference Andersen, Thorsen and Kousgaard2014).
Second, a study among hospital specialists presented concerns about stratification practices at the GP level (Munch et al., Reference Munch, Arreskov, Sperling, Overgaard, Knop, Visbøll and Røder2016a, Reference Munch, Bennich, Arreskov, Overgaard, Konradsen, Knop, Vilsbøll and Røder2016b). This is indicative of the professional cleavages that influence the interactions between hospitals and GPs. It also suggests that DMP implementation is dependent upon perceptions and normative structures among health care professionals.
4.4 Municipalities
Local government in Denmark is organized into 98 municipalities with an average of 59k inhabitants. The municipalities are governed by citizen-elected councils, and they have responsibility for a broad range of welfare services including home care, elderly care, health promotion, prevention and rehabilitation outside of hospitals. The municipalities are financed by local income taxes and block grants from the state and their budgets are subject to nationally imposed restrictions. This means that activities in health care, including new DMP-related activities, must be financed within the budget frame. At the same time, municipalities also pay a co-payment to the regions when patients are admitted to hospital. In principle, this gives them an incentive to develop prevention activities e.g. for diabetes patients. However, the uncertainties and time lag between interventions and effects tend to weaken this incentive (Pedersen, Reference Pedersen2019). Although the economic incentives for municipal engagement in DMP activities appear weak, they are also subject to other regulatory pressures through policy recommendations and negotiated agreements with the state, and mandatory health agreements with the regions.
The 98 municipalities are members of ‘Local Government Denmark’ (LGD), which acts as an interest organization and negotiates agreements with the state on behalf of the municipalities. LGD also participates in policy development fora with state and regional authorities including a working group established by the Danish Health Authority to develop recommendations for municipal rehabilitation and prevention efforts for chronic care patients (Danish Health Authority, 2016). In spite of these central recommendations, the municipalities have leeway to adjust to local conditions and preferences when designing health policy (Diderichsen et al., Reference Diderichsen, Scheele and Little2015; Holt et al., Reference Holt, Carey and Rod2018a, Reference Holt, Rod, Waldorff and Thomsen2018b).
Our municipality survey confirms that there is variability across municipalities. All municipalities offer an individual consultation at first referral as recommended in the guidelines published by the Danish Health Authority (Danish Health Authority, 2016). This is used to develop an individual action plan for participation in municipal services. Evaluation of psychosocial barriers to self-care is part of the plan. A little more than half of the municipalities (50–60%) also offer tests of physical abilities, quality of life and physiological parameters such as body mass index (BMI), HbA1c, etc. to establish a baseline before commencing municipal services.
About 80% of the municipalities state that they have implemented changes in patient education programmes in response to DMPs. About 50% have implemented changes in the way they approach ‘early detection’, ‘communication with GPs and hospitals’ and ‘rehabilitation’ in response to the DMP. Most municipalities (80%) offer group-based patient education/disease management courses to T2DB patients. Smoking cessation, alcohol and dietary counselling is offered at an individual level by almost 80% of the municipalities. E-learning tools and online courses are offered for chronic care patients generally by more than 80% of the municipalities. Around 60% of the municipalities also offer group-based counselling on physical activity and slightly less than 60% offer group-based training specifically to T2DB patients (55%) or on an individual basis (30%).
Approximately 50% of the municipalities respond that DMPs have significantly improved coordination, quality and prevention in the joint municipal, GP and hospital ambulatory management of diabetes, while 35% respond that DMPs have led to some improvement.
When asked about barriers against collaboration with GPs the municipalities point to ‘low priority from GPs’, ‘limited understanding of municipal practices and services’ and ‘difficulties in establishing personal contact’ as the main issues. Information and communication technology (ICT) is mentioned as a barrier for about 20% of the municipalities.
The survey results generally confirm the conclusions from a previous study based on qualitative data from four municipalities (Andersen et al., Reference Andersen, Thorsen and Kousgaard2014). This study found that the municipalities engage in disease-specific counselling, general information about life style and specific activities to support life style changes such as physical exercise, cooking classes, etc. All four municipalities found that the level of GP referrals was lower than expected, and they had initiated information campaigns to GPs or had information meetings, although with a limited response from GPs (Andersen et al., Reference Andersen, Thorsen and Kousgaard2014).
Overall, it seems that municipalities have indeed developed new activities, or have redirected or relabelled existing activities in response to DMPs, health agreements and broader policy pressures. This has happened in spite of weak short run economic incentives to undertake such activities. However, limited referrals and imperfect interaction with GPs have meant that the uptake in some cases was slower than expected. This may partly be explained by limited awareness among GPs and in some cases also a concern about the lack of evidence for municipal activities.
4.5 Implementation dimensions and status
The previous section presented specific implementation actions by key stakeholders as well as background information to understand their positions. Using this information as input, the following section presents a status of implementation using the four dimensions from Table 1.
5. Stratification
All regional DMP include descriptions of stratification principles. As described above they vary somewhat across the regions, but generally stratification is supposed to take place at initial diagnosis and at subsequent annual controls.
The standard diagnostic follows the WHO guidelines of HbA1c ≥48 mmol/mol (6.5 pct.) in two successive measurements. Patients are generally stratified into three levels: level 1 is well-controlled diabetes without complications. Treatment of these patients takes place in general practice. Level 2 is for diabetes patients with initial signs of complications or high risk of complications. Treatment in general practice is intensified for this group or they may shift to hospital ambulatories. Level 3 is complex diabetes with manifest late complications and co-morbidities (particularly cardiovascular disease (CVD)). Treatment takes place in specialized hospitals in most cases.
Municipalities cannot employ medical doctors and can, therefore, not assume formal responsibility for treatment of patients. However, all patients are to be offered a rehabilitation/self-management programme at the municipal level when possible. Some municipalities have established ‘health centres’ or ‘municipal clinics’ that can offer training/prevention and ongoing support to chronic care patients, while treatment responsibility remains with the GPs or hospital ambulatories.
It was initially expected that stratification would lead to a distribution with most patients in level 1, fewer in level 2 and least in level 3. However, previous Danish studies have shown that 61–75% of T2DB patients are stratified to level 2, while only 18–29% are stratified to level 1 (Rytter et al., Reference Rytter, Andersen, Friborg and Schroll2010). It has further been shown that most patients with T2DB are managed in general practice and not systematically according to the recommendations for their stratification level (Qvist et al., Reference Qvist, Glintborg, Andries and Hansen2008; Lassen, Reference Lassen2009; Drivsholm et al., Reference Drivsholm, Schultz-Larsen, Poulsen, Jensen, Jørgensen, Christensen, Munck and Bernsten2017). The critical issue is how to determine the right level of treatment for level 2 patients that may be subject to intensified treatment at the GP level, or shifted to hospital ambulatories. Several explanations have been offered for the actual stratification patterns (Munch et al., Reference Munch, Røder, Rasmussen, Middelfart, Kaarsberg, Bennich, Knop and Vilsbøll2018). First, it may be due to capacity problems in specialized hospital ambulatories. Second, it may be caused by limited knowledge and application of the risk stratification model in general practice (Gjessing et al., Reference Gjessing, Jørgensen, Møller, Huge, Dalgaard, Nielsen, Thomsen and Buch2014), which may imply that GPs underestimate the risks for level 2 patients and therefore do not refer enough patients to specialist ambulatories, or refer them too late. A study among hospital specialists support this picture as it indicates that the stratification levels of newly referred patients is too low (Munch et al., Reference Munch, Arreskov, Sperling, Overgaard, Knop, Visbøll and Røder2016a, Reference Munch, Bennich, Arreskov, Overgaard, Konradsen, Knop, Vilsbøll and Røder2016b). Third, it is possible that patients themselves prefer not to be treated in specialized ambulatories e.g. due to travel distance. In any case, it appears that there are ongoing tensions and ambiguities related to professional interpretations of which patients and how much responsibility for higher risk profiles the GPs can take on.
Preliminary results from a register-based study of diabetes patients in Denmark in 2015 illustrate some of the dilemmas (Pulleyblank et al., Reference Pulleyblank, Laudicella and Olsen2019). This study finds evidence that average hospital-monitored diabetes patients are in poorer health than GP-monitored patients, and the quality of health care delivered to hospital-monitored patients is higher. This is in line with the overall stratification principles. However, the same study also suggests that at least in 2015 there were possible net cost savings to be realized by moving care of some hospital-monitored patients to primary care. The critical issue is to identify those patients that can be transferred from hospital to GP management, and to ensure that the right patients are referred from GPs to specialists. The regions have attempted to deal with this by developing detailed guidelines, and requiring that GPs consult with a hospital specialist before referring. However, a clear standardization of how the criteria should be interpreted is complicated by the heterogeneity of the patient group with different risk profiles and needs e.g. according to age and personal resources.
All in all, it appears that there is some degree of persistent ambiguity and disagreement in the interpretation and implementation of the stratification model between GPs and hospitals in spite of attempts to clarify criteria.
6. New activities or changing practices
As presented above, several new activities have been implemented by regions, hospitals, GPs and municipalities in response to DMPs. The municipality survey illustrated that the municipalities generally had responded to the policy change by scaling up on their rehabilitation and patient-counselling activities, which have become more targeted and comprehensive. Most GPs have accepted the transition to the ‘chronic care management fee’, and appear to have stepped up their efforts to develop personalized diabetes management plans supported by digital data collection, which also supports quality development in cluster collaborations with other GPs. Overall, this has led to a reduction in the number of hospital referrals. This is also supported by regional requirements that GPs must consult hospital specialists before referring patients for treatment at that level.
In spite of these general indications of positive development at the GP level, there is variation in how many of the eligible patients that the GPs had connected to the ‘chronic care fee’ and how much they engage in associated activities. Half of the GPs had connected 30% or less of their eligible patients, while the lowest decile had connected 6% or less in 2018 (Sundhedsdatastyrelsen, 2020). Although this is the first year, and the overall enrollment is still increasing, it illustrates the variability in GP responses to general policy measures. This variability may partly explain the concerns raised from some hospital specialists about inadequacies in treatment and referral patterns from the GPs, and the municipal observations about variability in referrals to their chronic care services.
Hospital level changes include a number of different out-going projects as described above, and also some internal projects such as the establishment of internal diagnostic centres, cross-disciplinary teams and disease centred structures. These changes are parallel and sometimes prior to DMP implementation and illustrate that many other concurrent changes are taking place within chronic care management.
Both GPs and hospitals have been subjected to accreditation processes based on standards that make reference to both DMPs and clinical guidelines. There is no systematic evidence about the extent to which this has been a driver for changes in practice.
Joint municipal–regional instruments for improved coordination and patient journey management have been implemented across the country, albeit at different speeds and in different forms (Holt et al., Reference Holt, Carey and Rod2018a, Reference Holt, Rod, Waldorff and Thomsen2018b). Some examples were presented above, others include joint ‘health houses’ with municipal and regional staff, cross-disciplinary teams, cross-sector collaboration agreements, formal communication and collaboration models, designated contact persons with coordinating functions, local economic incentives and structurally supported coordination e.g. through shared employment, regular joint meetings and study visits with discussion of specific patient cases (Danish Health Authority, 2013). There is a lack of hard evidence about the effects of these activities, but respondents in five regions and 86 municipalities covered in a survey from 2013 generally report positive experiences, although the division of responsibilities continues to cause uncertainty (Danish Health Authority, 2013; Holt et al., Reference Holt, Rod, Waldorff and Thomsen2018b). Positive effects of individualized patient journey management are reported with regards to reduced dropout rates for rehabilitation, improved patient experiences and improved communication.
Temporary funding for implementation projects can be a problem for sustainability of new activities (Danish Health Authority, 2013). The lack of coordination with existing and parallel efforts is another issue that has been known to negatively impact motivation and good use of resources (Danish Health Authority, 2013; Holt et al., Reference Holt, Carey and Rod2018a). The issue of sustainability and coordination of efforts is also highlighted in a recent report commissioned by the Ministry of Health, Danish Regions and Local Government Denmark (the association for municipalities) (McKinsey and Company, 2017). The report confirms the observation that regions, hospitals and municipalities have initiated many activities to promote chronic care coordination including diabetes in the past decade. However, the lack of long-term funding and issues of scalability in the face of diverging incentives create challenges. A key concern here is that many project-based activities arise from individual and network-based entrepreneurs, which make them vulnerable for long-term sustainability (McKinsey and Company, 2017).
7. Shared governance infrastructure
Mandatory ‘health agreements’ are negotiated every 4 years between the regions and the municipalities in each region. The purpose is to establish shared visions and to coordinate roles and responsibilities with regards to patient pathways. DMPs are integrated into the health agreements, which establish the formal framework agreement for obligations and division of labour. Regular follow-up meetings assess the progress and provide fora to discuss coordination issues (Danish Health Authority, 2013; McKinsey and Company, 2017; Implement Consulting Group, 2011). In spite of the good intentions of the ‘health agreements’ it has been found, that implementation can be challenging (Vrangbæk et al., Reference Vrangbæk, Rommetvedt, Stigen and Opedal2014; Holt et al., Reference Holt, Carey and Rod2018a, Reference Holt, Rod, Waldorff and Thomsen2018b). The agreement process is criticized for being too ‘bureaucratic’ producing overly detailed plans, that appear less realistic for ‘street level bureaucrats’ in the various health care organizations (Vrangbæk et al., Reference Vrangbæk, Rommetvedt, Stigen and Opedal2014).
A critical issue is that GPs are supposed to comply with the regional health agreements, but they are not a formal partner of the agreement (Danish Health Authority, 2019). This tends to weaken motivation and makes enforcement difficult (Danish Health Authority, 2013; Vrangbæk et al., Reference Vrangbæk, Rommetvedt, Stigen and Opedal2014; Holt et al., Reference Holt, Carey and Rod2018a).
A number of clinical guidelines and recommendations have been developed within the area of diabetes management as described above. The regional DMPs make reference to these guidelines and both GPs and hospitals have been subjected to accreditation processes based on standards that also reference DMPs and clinical guidelines (e.g. Danish Endocrinological Society, 2011; Danish Health Authority, 2016, 2017a, 2017). There is limited evidence about the impact of this in practice, and the previous accreditation scheme has been replaced by a more flexible approach after critique from hospitals and health care professionals.
ITC may be used as a tool to facilitate shared governance and Denmark is often mentioned as a frontrunner in the use of IT and telemedicine in health care (Dobrev et al., Reference Dobrev, Haesner, Hüsing, Korte and Meyer2008; Economist, 2011; Frost and Sullivan, 2017). A number of telecare solutions are currently being implemented in addition to the existing facilities for sharing information as documented in a joint regional/national website called ‘Det telemedicinske landkort’. Evaluations show good results in several projects, although issues of integration into local practices and accept by health care professionals and patients persist (Danish Health Authority 2013; Høyer and Wadmann Reference Høyer and Wadmann2020; Juul Nielsen and Langstrup Reference Juul Nielsen, Langstrup, Mayer, Carpes and Knoblich2014).
Multimorbidity appears as a particular challenge, as the DMPs are developed for specific diseases while there is less evidence to support the organization of care for patients with complex care needs due to multiple conditions (Frølich et al., Reference Frølich, Jacobsen and Knai2015).
8. Alignment of incentives
Remuneration of regional hospitals, municipalities and GPs has developed over time according the institutional governance structure for each of the three domains, and in response to policy issues and specific focus points. Although this has supported policy goals within the three domains, it has proven much harder to find good solutions for activities such as DMPs that require collaborative efforts and shared responsibilities.
This is clearly illustrated in the productivity focus and activity-based remuneration scheme for hospitals which was introduced in response to long-waiting times and rising expenditures in the previous few decades. The schemes have led to a remarkable increase in productivity in the hospital sector, but have also created negative side effects in terms of unnecessary activities in some areas and disincentives to shift activities out of hospitals and to engage in telecare solutions (Højgaard et al., Reference Højgaard, Kjellberg and Bech2018). In response to such concerns, it was decided to remove the fixed requirement of a 2% annual productivity increase in 2019 and to convert the national the ‘activity funding pool’ for regions to a ‘proximity funding pool’ based on criteria that are aligned with DMP principles (see above). However, activity is still monitored, and the ‘proximity funding pool’ only covers a minor portion of the total financing of the regions.
GPs are remunerated according to tri-annual national agreements between the associations of GPs and the Danish Regions which determines fees and obligations to undertake new activities.
A lump sum annual chronic care management fee for patients with diabetes and/or COPD was introduced with the general agreement for GPs in 2018 (see details above). The scheme is linked to the ambition of moving responsibility for 25,000 chronic care patients from hospitals to GPs. There are no explicit incentives linked to quality, but GPs are expected to engage in data supported quality development activities and participate in cluster collaborations with other GPs. The fee is paid in monthly portions and is suspended if responsibility for treatment is transferred to hospitals i.e. due to a deterioration of the health status. This limits the economic incentive to refer patients unnecessarily, and this is further reinforced by easier access to guidance from hospital specialists in all regions, and requirement/strong encouragement to consult them before transferring responsibility (Forløbsbeskrivelse Region Midt, 2019).
Patients that achieve a stable HbA1c level can be maintained within the programme, and GPs maintain their fee. This gives the GPs an incentive to optimize management of chronic care patients as this presumably leads to fewer contacts.
Municipalities are mostly financed through block grants. As explained above this means that new or expanded activities must be funded within their budget giving little short-term incentive to engage in chronic care management activities. A co-payment for hospital admissions was introduced in 2008 to create incentives to increase prevention activities, but it has been argued that incentives are too weak as the benefits are uncertain and mostly realized in the long term (Pedersen, Reference Pedersen2019). Nevertheless, the co-payment reinforces the political incentives to respond to national policy pressures and to scale up prevention and rehabilitation for chronic care patients in the municipalities.
Danish municipalities are not allowed to employ medical doctors for treatment of patients and can therefore not assume formal responsibility for treatment of chronic care patients. However, they have important roles for prevention and rehabilitation, and since 2007 many municipalities have established temporary ‘acute beds’ relying on specialized nurses and collaboration with GPs. Many municipalities have also created ‘health centres’ or ‘health houses’ with co-location of municipal health staff, GPs and other health care professionals. The health centres usually provide support, counselling and rehabilitation. The municipal co-payment is a contributing factor behind these initiatives, but they are also spurred by the regional/municipal ‘health agreements’ and ‘political incentives’ to maintain legitimacy.
Previous studies have pointed to weak alignment of incentives for hospitals, GPs and municipalities as a barrier to the ambitions of improved coordination (Holt et al., Reference Holt, Rod, Waldorff and Thomsen2018b) and thereby also for the ambitions with DMPs. The presentation above underlines the lack of a joint incentive structure with direct mechanisms for sharing risks and rewards for improving quality and value for chronic care patients. The three main actors (regions/hospitals, GPs and municipalities) operate under different economic steering regimes, where hospitals and GPs until 2019 had incentives to increase activity, while municipalities had disincentives to do so. The municipal co-payment, the GP chronic care fee from 2018 and the 2019 conversion of the activity-based funding pool for regions to ‘proximity funding’ partially addresses these issues. A more comprehensive solution might become part of an upcoming reform initiative in 2021. Table 2 summarizes the main observations about implementation.
Summary of implementation dimensions
9. Conclusion and discussion
The implementation analysis presents a picture of a long and gradual process from the initial formulation of DMP principles in 2008 to the stepwise uptake in regional, municipal and GP governance frameworks and practices. This is not surprising in light of classical implementation studies (Pressman and Wildawsky, Reference Pressman and Wildawsky1973) and previous analyses of DMP implementation (Nolte et al., Reference Nolte, Knai and McKee2008). However, it should also be noted, that although the process has been slow and with occasional halts and even reversals, there has been an overall momentum in developing initiatives to transform chronic care in accordance with the broad DMP principles. This has led to a more structured approach to chronic care, with increasing emphasis on the role of GPs and municipalities and more formalized approaches to coordination, personalized management and quality development. The developments are supported by political incentives, regulatory measures and some economic incentives, while an integrated economic incentive structure for chronic care management across governance levels has not yet been developed.
Although there is positive momentum at the general level, a number of areas indicate continued challenges in the efforts to structure the complex reality of chronic care. Actors that are furthest from the decision apexes are least likely to be aware of the formal content of DMP policies, and there is heterogeneity in the practices among municipalities and GPs. This can only partially be justified by the need to tailor activities to local needs and preferences.
Insufficient knowledge about activities and competencies at the other governance levels appear to create inefficiencies, e.g. in referrals to municipal services and in the stratification of patients between GPs and hospitals. Stratification ambiguities are also resulting from the heterogeneity of the patient population with different risk profiles and co-morbidities.
Digital solutions are widely applied for communication, and increasingly also as tools to collect data and support disease management. Yet, the results do not always match the aspirations as they are implemented in real-life health care settings and in patient populations with different levels of readiness and accept.
The overall observations about general momentum, but stepwise and partial implementation align with previous international studies from Germany and Austria (Schang et al., Reference Schang, Thomson and Czypionka2016) and internationally (Felder, Reference Felder2006; Ham Reference Ham2010; Miksch et al., Reference Miksch, Triecshmann, Ose, Rölz, Heiderhoff and Szecsenyi2011; Nolte and Hinrichs, Reference Nolte and Hinrichs2012).
One might have expected that the smaller size and the public integrated nature of the Danish health system could have facilitated a faster and more homogenous implementation. Reasons for not achieving this can be found in the multilevel governance arrangements, where national level health policies must be implemented in an interactive process involving regions, hospitals, GPs and municipalities. This is time consuming and may lead to variation in practices and hindrances to implementation. On the contrary, there are benefits of the negotiated multilevel scheme in terms of securing involvement by relevant stakeholders in developing realistic solutions through bottom-up processes (Kaehne, Reference Kaehne2019). This may partly explain that in spite of the implementation concerns outlined above there are indications that diabetes care results have improved in Denmark over the past two decades in terms of lower mortality (Gaede et al., Reference Gaede, Lund-Andersen, Parving and Pedersen2008; Jorgensen et al., Reference Jorgensen, Almdal and Carstensen2013; Faerch et al., Reference Faerch, Carstensen, Almdal and Jorgensen2014) and fewer severe complications (Andresdottir et al., Reference Andresdottir, Jensen, Carstensen, Parving, Rossing, Hansen and Rossing2014; Jorgensen et al., Reference Jorgensen, Almdal and Faerch2014). The Danish case has thus illustrated, that even in a public integrated health care system it takes time and persistent efforts to implement complex policy changes. Stakeholders engage in processes of reinterpretation and renegotiation, which lead to recursiveness and variability in the chain from policy formulation to practice. Such dynamics are likely to also apply in other countries implementing complex health care programmes, while the stakeholder configurations and the specific pathways vary. It is, therefore, important to take a systemic approach to understand multilevel implementation processes in the efforts to structure complexity in health care.
