Implementing pay-for-performance in primary health care: the role of institutional entrepreneurs

Abstract

Institutional entrepreneurs are vital for facilitating non-incremental health policy change in complex institutional settings where established traditions and practices carry considerable weight. This paper describes a comparative case study of health policy-making which shows that Kingdon’s Multiple Streams Framework for non-incremental policy change requires enhancement to explain results in policy-making in two Westminster unitary majoritarian jurisdictions. The most similar systems comparative study found that historical, rational choice, organisational and discursive institutionalist approaches explained the policy change and variation observed better than agency-based approaches did. However, institutional entrepreneurs were important in both cases. Differences in coordinative discourse help to explain the differences in degree of change achieved in each case study and highlight the importance of discursive institutionalist approaches in bridging institutional and agency-based approaches.

Keywords:

• Institutional entrepreneurs
• Multiple Streams Framework
• policy change

Introduction

How important is agency in policy change? The research reported here is a comparative case study which addresses national-level health policy-making in England and New Zealand (NZ). It assesses the utility of both Kingdon’s multi-theoretic Multiple Streams Framework (MSF) (Kingdon, 2010) and other explanations for non-incremental policy change and variation (John, 1998). This research adds to the burgeoning literature applying and critiquing the MSF (Jones et al., 2016) and finds patterns of change which run counter to Kingdon’s predictions that non-incremental change occurs in conditions of ambiguity, fluid participation of actors and unclear technology for implementation.

The research makes a distinctive contribution to this debate about the utility of the MSF through its use of an integrated analytical framework which includes the MSF and other explanations for policy change. Using this framework, which synthesises several theories of policy change into a coherent approach to assessment of policy change and variation, it considers two comparable episodes of policy change within similar structural settings and well-established institutional settings. This allows for analytic generalisation (Yin, 2009). The analytical framework uses the MSF, as interpreted and adapted by Zahariadis (2007), as well as an analysis of institutional, socio-economic, ideational, network-based and rational actor explanations for policy change, in each case. Then the two cases studies are compared using an attribution process described by Castles (1991) to ‘locate some particular features in which otherwise very similar nations differ (so that) we are entitled to suggest it is attributable to one of the few other factors distinguishing them’ (p. 5).

The findings of this analysis show that institutional explanations are most persuasive in these two episodes of policy change. However, institutional entrepreneurs (Campbell, 2004; Crouch, 2005; Tuohy, 2012a) were actively engaged by the state and used different coordinative discourses, or interactive processes of conveying ideas (Schmidt, 2008), with differing effects. Institutions and agents were both important in explaining the policy change and variation observed, lending support to writers who seek to bridge agency-based and institutional theory and supporting calls (Bakir, 2009) for an integrated framework for analytical interpretations of the logics of human action which incorporate structural, institutional and agency-based explanations for policy change (Bakir, 2017).

The analytical framework used here is original and one of a smaller number of studies which are hypothesis-driven (Cairney & Jones, 2016). In common with other writers (Bakir, 2009; John, 1998; Mucciaroni, 1992; Schlager, 2007), recommendations for the enhancement of the MSF are made, specifically to incorporate historical, institutional and longitudinal factors in non-incremental policy change into the MSF.

In the first section, the MSF, literature utilising the MSF, and institutionalist approaches are briefly reviewed. The policy context, research methods and the analytical framework are described. In the second section, the two case studies are presented. Findings are presented and finally, the conclusion addresses the need to enhance the MSF and to use four approaches to institutionalist analysis to make sense of the policy change and variation observed in the two case studies.

Theory and methods for the research

The Multiple Streams Framework

John Kingdon’s MSF was first published in 1984 and has been described as one of the most important contributions to ‘an adequate theory of public policy’ (John, 1998, p. 173). Currently the MSF is one of the most prolific and widely recognised of the approaches to public policy-making utilised by scholars (Jones et al., 2016). Kingdon’s MSF places the individual actor as a policy entrepreneur at the centre of non-incremental processes of public policy-making (2010). Challenging the prevailing view that policy-making was usually rational and incremental (that is, proceeding in small steps rather than big changes), Kingdon predicted that major policy change needed circumstances of undecided or ambiguous policy preferences, fluid participation in the policy-making process (or allowing many actors to participate), no clear technology to solve problems and the intervention of an exogenous policy entrepreneur. Such actors seek chances to couple or link existing and known policy problems, political will, and available policy solutions into credible policy-making proposals. They must take advantage of fleeting windows of political opportunity to successfully place particular policies onto a politician’s agenda for action.

The MSF has been applied in health policy settings outside the United States by several authors, as with the comparative case study presented here. For instance, Strand has applied the MSF to health policy change addressing health inequalities in a small single case study in Norway and demonstrates the importance of a policy window occurring at a change of government which enables a non-linear shift (Strand & Fosse, 2011). Kusi-Ampofo et al. applies the MSF to a Type 1 case study of non-incremental health policy-making in a developing country, studying a change in funding arrangements which was also triggered by the policy window of a change of government, using a policy evolution analysis method (Kusi-Ampofo, Church, Conteh, & Heinmiller, 2015). Smith considers the role of entrepreneurs in a process change inside a tertiary care facility (Smith et al., 2016) and finds the role of physician entrepreneurs important to the process of policy change.

Much research has focused on Kingdon’s view of the policy entrepreneur. Oliver and Paul-Shaheen concluded that skilled and committed leadership rather than the skills of a solo operator was a vital attribute of policy entrepreneurship. Importantly, these researchers identified that key institutional factors in major reform and innovation were the resources, structure and culture of the health policy community in each state (Oliver & Paul-Shaheen, 1997, p. 739). Mintrom further extended our understanding of the attributes of policy entrepreneurs, finding that social acuity, ability to define policy problems, building teams, and exercising leadership were characteristics of policy entrepreneurs (Mintrom & Norman, 2009). Deep knowledge of relevant procedures and local norms within institutions, which Mintrom called insider sensibilities, can significantly increase the ability of actors to instigate change.

Although the studies differ widely in their scale and methods, they confirm the transportability of the MSF beyond the United States and its utility and relevance for describing policy change in a variety of settings.

Institutionalist approaches to decision-making

Institution-based rather than agency-based theories for policy change also offer explanations for policy change and variation. Institutions provide a structure for individuals and organisations to interact in and also create the incentives which influence actors’ choices (Sabatier, 2007, p. 309). Historical institutionalist approaches see decision-making reflecting historical experience arising from complex struggles and bargaining amongst organised groups, which have a continuing effect (Campbell, 2004, p. 25). Rational choice or ‘consequentialist’ institutional frameworks (Campbell, 2004, p. 10; March & Olsen, 2008) assume actors make self-interested choices. Organisational institutionalist approaches (Campbell, 2004, p. 17) assume actors follow rigorous rules and professional standards. Schmidt offers a fourth, ‘discursive’ institutionalist approach which has considerable relevance for the two case studies reported here (Schmidt, 2008; Winkel & Leipold, 2016). In this approach, the representation of ideas (how agents say what they are thinking about) and the way actors generate and communicate ideas, for instance in a coordinative process amongst policy actors (to whom they say it), within given institutional contexts (where and when they say it) are relevant. Actors achieve ‘discursive agency’, defined as ‘an actor’s ability to make him/herself a relevant agent in a particular discourse, by constantly making choices about whether, when and how to identify with a particular subject position in specific story lines within this discourse’(Winkel & Leipold, 2016, p. 112).

There is much dispute between theorists about the strengths and weaknesses of the four institutionalist approaches. This has resulted in calls for a more integrated approach, drawing on all four paradigms to improve the analytic power of institutionalist approaches, especially for understanding change (Bakir, 2009; Campbell, 2004) and to link these to agency-based theories of policy change.

Institutional entrepreneurs

Research in the institutionalist tradition has identified ‘institutional’ rather than ‘policy’ entrepreneurs to be very important participants in policy-making processes in Westminster unitary majoritarian jurisdictions (Campbell, 2004; Crouch, 2005; Tuohy, 2010, 2012a) These actors will typically straddle two worlds to provide a bridge between the state and their profession or sector, termed boundary-spanning activities’, which are often mandated by public officials. To do so, they place their specialised knowledge and/or other private resources such as capital or technology (Tuohy, 2012a, p. 3) at the service of the state. Campbell has described this as a process to recombine elements in a repertoire through a process of bricolage from which new institutions are created, different from but resembling old ones (Campbell, 2004).

Crouch hypothesises that ‘to carry out this kind of analysis we must have an approach that can perceive two or more, or fragments of several … institutional forms coexisting within one political economy’. This institutional heterogeneity facilitates innovation by presenting actors with alternative paths (Crouch, 2005, p. 150). In their actions, institutional entrepreneurs are helping to forge new institutional arrangements or to ‘scale-up innovations’ for national utilisation. Tuohy found that health policy-making, including in Britain, between 1990 and 2010 occurred in the context of market-oriented reforms which had both ‘favoured and been accelerated by the emergence of institutional entrepreneurs’ (Tuohy, 2012a, pp. 2–4). Institutional fragmentation, heterogeneity (and looseness of coupling of resources in the environment to permit reconfiguration) and political uncertainty (created for instance by periods of major policy change) provided favourable conditions for institutional entrepreneurs. Tuohy reinforces the elements of risk and recombination as critical to the definition of entrepreneurship.

Integrating structural, institutional and agency-based approaches

Bakir (2013) explores the interaction of structure, institutions and agency and utilises Kingdon’s MSF in a case study of non-incremental change in monetary governance which highlights the role of policy and institutional entrepreneurs. Bakir proposes a framework in which multiple interactions among structural, institutional, and agency variables drive policy change. His research emphasises the importance of the discourse of endogenous entrepreneurial actors in mediating ideas, as well as vested interests, in effecting institutional change. Bakir identifies the need to expand Kingdon’s MSF by adding an historical perspective, by finding the entrepreneurial role relevant at many levels as well as the national level and by extending this role beyond innovation into many stages of policy design and implementation.

Methods

In this research the process of similar episodes of health policy-making between 2001 and 2007 in two Westminster political jurisdictions, England and New Zealand (NZ), was described. The utility of Kingdon’s MSF, as enhanced by Zahariadis’ description of its elements and sub-elements, in explaining these two processes of policy-making was then analysed. In particular, the research considered whether policy entrepreneurs were the drivers of non-incremental policy change in the two contemporaneous policy-making episodes. Kingdon’s definition of non-incremental change was used, namely that a policy idea reaches the agenda and dominates it in a visibly significant way over a four-year period (Kingdon, 2010). The research also explored all other potential drivers of non-incremental policy change as set out by John (1998), including institutions, networks, socio-economic circumstances, ideas, and rational actors in each policy-making episode.

A comparative case study methodology in a most similar systems design (Yin, 2009) was used, based on purposeful selection of the two case studies of contemporaneous implementation of pay-for-performance within general practice funding in two countries, which showed outputs which differed in size, scope, speed of implementation and impact on health outcomes. These differing results occurred in jurisdictions which displayed similar health systems, trajectories of health system development, and political systems. Three research questions were considered:

•	In what aspects and why did two similar episodes of policy formulation and implementation in two similar jurisdictions follow different processes and have different outcomes?
•	How well do the elements of Kingdon’s MSF describe and/or explain what happened at each stage of the policy-making process?
•	What new relationships between variables can be identified from the analysis which may enhance or extend Kingdon’s MSF?

The unit of analysis was the process of policy design of the pay-for-performance component of primary health care policy changes in each country. The focus was on the processes of agenda setting and alternative selection leading to policy formulation and authoritative decision (Kingdon, 2010). However, evidence from the ten-year period before and after the policy-making period and evaluations of the two schemes was also considered.

A qualitative methodology was adopted, utilising documentary analysis and seeking semi-structured interviews with 26 proximate decision-makers, leaders, and participants who were directly engaged in the design of the two pay-for-performance systems. Ethics approval was gained from the relevant bodies in both England and NZ to interview people identified by their managing organisations as having responsibility for negotiating the terms and conditions of the pay for performance schemes. They were invited to participate in a one hour interview and asked about their role in the process of policy design and/or implementation, their perception of the dynamics in the process of design, what their expectations about the process had been and to what extent these were realised.

Interviews were transcribed immediately, and the process of interviewing participants was continued until no new data were appearing and saturation was achieved. Analysis of the transcribed interviews proceeded inductively, involving reading and rereading the data and note taking (rather than coding) of themes, looking for patterns in the data and the prevalence and strength of themes (Braun & Clarke, 2006). A comprehensive description was developed of the processes of policy-making from the point of view of each participant (who did what, when, how and why). This was cross-checked and corroborated against the data collected from other participants and from documentary evidence available to the researcher, such as government reports and media reports, to validate the causes, consequences and relationships which appeared in the data. Then tables setting out patterns of similarity and difference in descriptions of processes by role of participant (such as politician, doctor, civil servant) were developed and considered, in order to understand whether there was evidence of common role-based expression of interests or understandings. The data corpus was then written up as two case study narratives which described what happened, why and how this was perceived by different participants. These case studies were then re-submitted as summaries to all interviewees from each country case study for verification, cross examination and testing of trustworthiness of the analysis. These summaries included the data induced by the researcher through thematic analysis, as well as extensive descriptive material drawn directly from participants, in a process to seek validation of these conclusions. All participants responded and minor changes were made to the analysis following this process.

As part of each case study, structural and historical contexts were described and points of structural similarity between both countries noted, such as their political and health systems. Points of historical divergence, such as the relationship between state funders and general practitioners, were also noted. Then, for each case study, the fit of each process of policy-making with the five major elements and the sub-elements of Kingdon’s MSF, as interpreted and adapted by Zahariadis (2007), was analysed and described. The analysis focused on the extent to which there was evidence of non-incremental change and whether it was accompanied by ambiguity, fluid participation, unclear technology, if there was a key role for policy entrepreneurs and the nature of the policy window. Next the fit with causal explanations of policy change and variation in the institutional, network-based, socio-economic, ideational and rational actor-based literatures was applied to each case and written up. These cases were then compared using a cross-case synthesis technique (Yin, 2009, p. 156) to develop a set of similarities and differences and apply Castles’ attribution process (Castles, 1991). In this process of analysis, the particular features in which the two case studies differed were considered as drivers of the difference in policy-making outcomes observed in the two case studies.

The two case studies

Contextual considerations for the two case studies

Between 2001 and 2007, England and NZ implemented national pay-for-performance schemes within their primary health care sectors in order to increase the influence of the state over general practice activities and improve health outcomes for citizens. Both countries are Western majoritarian unitary democracies with strongly adversarial political systems and high levels of autonomy and authority for central government (Pollitt, Harrison, Dowswell, Jerak-Zuiderent, & Bal, 2010; Richards & Smith, 2002; Shaw & Eichbaum, 2008). Both had followed similar patterns of national health system establishment in the 1930s and 40s (Bolitho, 1984; Fougere, 1993; Ham, 1992; Hanson, 1980; Hay, 1989; Klein, 2006; Lovell-Smith, 1966; Tuohy, 1999). Thus, the structures, or broader contexts for policy-making, in each country are strikingly similar. In the literature on health system typologies, they are both assessed as being national health systems with comprehensive universally available health services, largely publicly financed through taxation (Burau & Blank, 2006; Scott, 2001). These shared features make them suitable for a most similar systems comparative case study method (Yin, 2009).

Both countries had then undertaken similar New Public Management (NPM)-inspired reforms to introduce competitive and market-oriented approaches into their health systems in the 1990s, driven by concerns about escalating medical costs (Davis & Ashton, 2000; Pollitt & Bouckaert, 2011). Tuohy’s description, above, of the institutional fragmentation and opportunities for system reconfiguration in England between 1990 and 2010 is an apt description also of the turbulence within the NZ health system during the same period (Davis & Ashton, 2000). Policy change was needed to increase preventive and population-based practice amongst general practitioners and resolve increasingly visible problems of variation in quality and access to primary health care leading to substantial disparities in health outcomes for some citizens. The highly individualised and treatment-centred practice norms of the general practice profession meant that general practitioners, especially in NZ, ‘had their strongest focus on patients who walked in the door … [and] did not understand ‘disparity’ in access to health services very well, feeling little responsibility for a population-based approach to health’ (O’Malley, 2003, p. 3). The two national pay-for-performance schemes were introduced as part of major institutional reconfigurations of each health system to require primary care organisations proactively to manage the health of a local enrolled population. Each government, contemporaneously, decided to make all or part of new money for general practice conditional upon doctors meeting new targets for the use of treatment guidelines and introduction of processes to practice population-based health care.

Such developments represented a challenge to the set of institutions which permit professional self-regulation and professional autonomy within medicine. They encroached upon doctors’ highly valued professional independence (Burau & Vranbaek, 2008; Freidson, 2001; Immergut, 1990). Medical institutions would see pay-for-performance is part of a set of ‘managerial notions that efficiency is gained from minimising discretion’ (Freidson, 2001, p. 3) and as the standardisation and commodification of care based on targets set not by the profession but by the funder, therefore inimical to these norms.

England: the design of the Quality and Outcomes Framework

In England in 2000, the pay-for-performance programme was part of a new NHS Plan (Secretary of State for Health, 2000) ‘to introduce systems where the money spent was linked to performance and where the service user was in the driver’s seat’ (Blair, 2010, p. 262) and to reduce health inequalities (Comptroller & Auditor General, 2010) by increasing the quality of care and introducing more preventive practices into primary care services. Both goals were in Labour Party manifestos and ministerial speeches during the election campaign in 2001. Once elected the manifesto promises were immediately implemented. The health policy adviser to the Prime Minister described the reform process as ‘constructive discomfort’ to put some pressure on professional autonomy within the medical profession by using top-down imposition of standards and targets and subjecting the profession to competition from other suppliers of medical services. In this reform programme, more money for general practitioners was conditional upon better performance: ‘GPs’ new contracts will allow them to earn around a third more, linked to markers of quality’ (Stevens, 2004, p. 41). The British Medical Association (BMA) represented all English general practitioners and was the sole bargaining agent on their behalf. BMA negotiators readily agreed that a pay-for-performance mechanism could form a major part of a new contract. However, they had little choice. It was hard for them to resist the strong expectations of the Secretary of State for Health that new money for general practitioners would be subject to pay-for-performance. A participant in the negotiations saw him as insistent that ‘there would be no pay rise for work already being done.’ He had a ‘bloody-minded determination for performance pay’ (Interviewee #8, London, September 2009).

Both parties then sought a large scheme. On the government side it was believed that the success of early pay-for-performance schemes (Spooner, Chapple, & Roland, 2000, 2001) had demonstrated the effectiveness of rewarding doctors for preventive practice, justifying a scheme with as many indicators as possible. For the BMA, the larger the scheme, the more money was available to their members. In addition, the pre-eminent role for the BMA in its relationship with government was at stake. If the BMA could not broker a popular and lucrative national deal for their members, it was said that the BMA feared other forms of localised contracting would erode their sole bargaining rights for all general practitioners.

Medical professionals dominated the membership of the teams which designed the pay-for-performance component of the new contract, the Quality and Outcomes Framework (QOF), constituting 7 of 11 members. An academic team was recruited to chair debates to consider evidence for suitable clinical targets for the scheme. Members of the different teams involved had known one another for a number of years in some cases. Almost all were practising general practitioners for at least a small part of their working week even when the remainder of the week was spent as a medico-politician or academic. A large scheme which included 146 targets and determined the level of over 30% of the new income of general practitioners, was jointly designed and negotiated by the government and BMA teams. The QOF could not be implemented without the design of a major new software application to extract and verify performance data from every general practice. This presented significant practical problems. There were many different suppliers of computer systems for general practice, some practices not yet computerised and concerns about confidentiality of patient information. Yet this project was achieved in 26 weeks. The scheme was implemented remarkably quickly by 99% of general practitioners. Within a year of its launch, performance against the targets was higher than expected and payments to general practices under the QOF exceeded budget allocations. Evaluations of the success of this large pay-for-performance scheme have been mixed and widely reported (Comptroller & Auditor General, 2008; Doran & Roland, 2010). A key study found that there were statistically significant associations between higher levels of achievement on QOF clinical indicators for coronary heart disease, hypertension, congestive heart failure, diabetes and chronic obstructive pulmonary disease and reductions in rates of ambulatory-sensitive hospital admissions for those conditions (Dixon et al., 2010).

NZ: the design of the NZ Performance Programme

The NZ Labour Party manifesto of 1999 also promised a new focus on improving the quality of primary care and increasing population-based and preventive health care. The health system was seen as ‘too focused on treatment services at the expense of improving the health of the community’ and an undertaking was made to contribute to significantly reducing disparities in Maori health status by comparison with other New Zealanders (NZLP, 1999, pp. 2 and 8). The focus on population-based and preventive health care grew out of concerns about disparities in health outcomes within NZ. These included a nine-year gap in life expectancy between Māori and non-Māori New Zealanders and between males living in the most deprived and least deprived geographical areas (Crampton, 2001). Once elected, Labour introduced a new network of Primary Health Organisations (PHOs) to manage primary care including general practice and changed the funding mechanism from fee-for-service to capitation. All New Zealanders were required to enrol through their general practice onto a PHO register for health care. Targets were set for PHO delivery of preventive as well as curative services (King, 2001).

Unlike England’s single payer and single ownership model for general practice, NZ had many types of general practice service delivery approaches and ownership forms, multiple payers for general practice services and a wide variety of interest groupings within the heterogeneous primary care sector. Two networks dominated the arena. For-profit Independent Practitioners’ Associations (IPAs) represented over 75% of NZ general practitioners (Malcolm & Mays, 1999). Most of these held contracts with the state to manage and improve the quality of prescribing and referral services. For IPAs, the reforms meant the end of these lucrative contracts, which had funded quality improvement activities amongst their members for many years. For this network, the reforms created tensions. Participants interviewed for this research reported the feeling amongst general practitioners at the time that the Labour party was anti-general practitioner. A second large grouping of not-for-profit community-governed health centres in New Zealand were concerned to improve access and equity of outcomes for poorer communities and fully supported the reforms. No single organisation existed to represent all general practitioners in their dealings with state funders.

The changes to primary care governance were rapidly implemented by the Labour government, despite misgivings in large parts of the primary health care sector. As part of this process, officials recommended the implementation of a national pay-for-performance scheme to improve the quality and equity of pharmaceutical prescribing and referrals to services within a population-based funding framework, now that budget management contracts with IPAs for this purpose had been cancelled. As in England, the idea drew upon successful local initiatives to incentivise performance which had been developed in the primary health care sector in previous years.

A working group of primary care stakeholders was convened to design the new Performance Programme (PP) for primary care. Although one of the Joint Chairs of the group was a general practitioner, general practitioners were a small minority in the group, invited as individuals with no representative mandate for their profession. These general practitioner members also held differing views about the role of general practice, reflective of the professional divisions in the country at large: some for and some against a state-led pay-for-performance scheme. The group utilised a variety of consultative methods and the assistance of academics to complete the selection of a small set of thirteen indicators for the PP. The choice of indicators was based pragmatically upon data elements already available from central sources even though they did not relate to many of the major health outcomes the Ministry was most interested in improving. This was because efforts to gain access to data within practice management systems, unlike in England, were opposed by most NZ general practitioners.

The set of indicators and the funding framework were approved in July 2005. By 2007, 81 of 82 PHOs had joined the scheme and, if eligible to receive payments, distributed all or some of these to practices. Achievement levels against the indicators averaged 81% in 2009 and the budgeted funding for the PP was never fully utilised.

Evaluations found that general practitioners had mixed views about the effectiveness of the PP, it had low visibility amongst clinicians, and it did not lift clinical quality as much as expected (Martin Jenkins & Associates Ltd, 2008). A later evaluation did however find that there was a statistically significant relationship between achievements under the PP for one of the thirteen indicators – immunisation of under two years olds – and vaccine-preventable ambulatory sensitive hospital admissions (Cranleigh Health, 2012).

The importance of agency

In both case studies, the research found that agency, and specifically entrepreneurial actors, played an important role in policy-making. Firstly, politicians in both countries recruited a small number of people with high team-building skills and high personal and professional credibility to be advocates, facilitators, and leaders of the negotiations on policy change. These actors were responsible for the nature and the quality of the discourse of policy design, reflecting the requirement of Ministers for a consensual approach to be taken to policy development. In England, Ministers chose as leader of their negotiation team Mr. Mike Farrar, someone who a participant confirms was ‘widely respected by Ministers, by civil servants, by managers and the profession’ (Interviewee # 3, London, March 2010). The BMA said

we got on well with [him] basically because of his palpable honesty really. He was a pleasure to do business with even when he was giving tough messages [and] would focus on the problems not the person. (Interviewee # 8 London, September 2009)

Participants for the research in England confirm that he utilised qualities of social acuity, problem definition, team building and leading by example, such as are identified by Mintrom (Mintrom & Norman, 2009). The freshness of this approach, in contrast to remembered and more traditional adversarial negotiating styles, was remarked upon and endorsed by all participants. In NZ, officials also deliberately recruited a respected general practitioner as a senior adviser within the Ministry of Health to support implementation of the primary health care reforms. He was also selected to co-lead the working group which designed the PP. In these roles, these actors resemble ‘policy’ rather than institutional entrepreneurs as defined by Mintrom and Oliver above but, being recruited by state actors subsequent to agenda-setting, did not operate in Kingdon’s classic model of the exogenous policy entrepreneur. In both case studies, the entrepreneurial skills exhibited are those which bring disparate ideas together, explore common ground and conciliate different interests.

Second, individual general practitioners who believed in pay-for-performance as a mechanism for improving quality of care within the profession were recruited by ministers or officials in both countries to champion pay-for-performance in negotiations with their professional colleagues. They fulfil Tuohy’s description of institutional entrepreneurs as they took personal and professional risks in doing so and in working so closely with politicians and public servants. They carried out this public mandate, encouraging change within the primary care service landscape, by combining the authority of the state with their specialised knowledge (Tuohy, 2012a). The Chair of the Quality Sub Group which designed the QOF in the English case study and the Co-leader of the working group in New Zealand, who is also referred to above as a policy entrepreneur, meet the definition of institutional entrepreneurs. Both had previous experience of the implementation of pay-for-performance-based quality improvement schemes and brought this experience into the heart of their respective national policy-making processes, effectively preparing the way for the next major episode of change (Tuohy, 2012b).

However, the nature of discourse differed substantially in the two case studies. Within the negotiating teams in England, debates were conducted according to familiar collegial, peer-to-peer approaches. A general practitioner negotiator for the government side described it thus:

When I was involved in the negotiation it felt like a practice meeting … we thought the patients were going to benefit … we were negotiating this in order to achieve patient benefit. Of all the team that was what we were genuinely trying to do. (Interviewee # 7, Crewe, October, 2010)

Another described the process as discussions rather than negotiations, conducted between peers with a shared purpose.

In the pay-for-performance policy design forums in NZ, general practitioners’ voices were to some extent crowded out by those of other primary health care professionals. A general practitioner participant in the policy design process described it thus:

This was a state-directed programme. I have often reflected that I don’t think a single thing [some general practitioner participants] said … was reflected in the programme that was rolled out … [such as] peer-led, based on feedback and performance data to individuals, the data referenced to colleagues and the group as a whole and using clinical meetings based on the evidence and outlier management visit … a non-judgemental peer accountability process. (Interviewee # 17, Wellington, December 2010)

Findings

Utility of the MSF for Westminster majoritarian unitary jurisdictions

The comparative analysis of the case studies showed that politicians in Westminster systems can successfully plan and implement non-incremental change without the conditions specified by Kingdon of ambiguity, fluid participation of actors, and unclear technology for implementation. Structural features, ideas and socio-economic drivers were well matched in both countries. Certain institutional features which differed between the two countries, rather than exogenous policy entrepreneurs, were the primary drivers of the scale of change and therefore outcomes observed, which differed between the two countries. In particular, the use of bargaining and negotiation processes enabled the large size, scale, and speed of design and implementation of the QOF in England and therefore the level of health outcomes attributable to the scheme. This differentiates the two episodes of policy formulation and implementation in two similar jurisdictions most strikingly.

Differences in the institutional framework within the general practice sub-system in each country are the primary driver of policy variation, facilitating change in England but frustrating it in NZ. In each country, institutional forms had arisen from highly path-dependent patterns of policy-making over successive decades. These had given rise over time to effective mechanisms for collective action and a well-developed working relationship with the state for English general practitioners. The way in which actors were able to represent their ideas and to whom they say it (through democratic processes within the BMA) enabled a sense of participation in the process which helped the scheme in England to succeed, in contrast to the scheme in NZ. Here, there was a poor and conflict-dominated relationship with the state and ineffective mechanisms for professional debate. Chief amongst these is NZ’s lack of a mechanism for the general practice profession as a whole to be represented by a bargaining agent and therefore to negotiate new policy proposals directly with state funders. The mandate held by the BMA in England to be sole bargaining agent for general practitioners holding general medical services contracts meant that general practitioners had trusted delegates representing them in their dealings with the state. Their own well organised professional forums enabled them to have a voice in discussions about the design of the pay-for-performance scheme.

There was also a reduced level of incentive for rational choice drivers to influence NZ general practitioners. They did not depend on the state for their income and could simply raise their fees to patients if they needed to. Unlike the English general practitioners who readily agreed to permit the sharing of their practice data with the state in order to increase the size of the scheme and the speed of assessment and payment of performance incentives, the NZ general practitioners declined to do so. This impacted on the size of the scheme, the type of targets which could be set and therefore the level of take-up of the new pay-for-performance policy. In both countries, the offer of a national pay-for-performance scheme was a lucrative inducement to general practitioners to increase preventive and population-based health actions in their medical practice. But in New Zealand, the benefits and opportunities of the proposed policy did not outweigh the perceived challenges it represented to general practitioners’ professional dominance and associated autonomy, monopoly and right of self-regulation (Freidson, 2001).

The process in NZ was perceived as a threat to professional organisational rules and standards and reduced the willingness, collectively and individually, amongst many NZ general practitioners to engage in the policymaking process. Rather than bargaining amongst general practitioners as in England, in NZ it was a process of broad-based primary care consultation. By contrast in England it was seen to be consistent with general practice norms. A participant captured the difference thus: ‘It’s curious the language countries use – in 1996 I went to London and the Secretary of State [for Health] was on stage talking and he was entirely comfortable with talking about a GP-led system. In this decade, the [NZ] Labour government couldn’t say the word general practice. In [the Ministry of Health] people still say primary care when actually they mean general practice’. (Interviewee # 17, Wellington, December 2010)

Entrepreneurial actors played an important role at the alternative selection and implementation stages of policy-making, using coordinative discourse and serving as catalysts for change in both case studies. In Schmidt’s words, ‘discourse puts the agency back into institutional change by explaining the dynamics of change in structures through constructive discourse about ideas’ (Schmidt, 2008, p. 316). These entrepreneurial actors operated at the direction of each government to lead or encourage change in institutional and governance settings within the medical profession. The roles played by these actors in policy-making in both countries arguably made the difference between policy change which was consensual and that which would have had to be enforced upon an unwilling medical profession. These policies would have been implemented without the roles played by these entrepreneurs. But these particular types of actors complemented and supported the institutional drivers of policy change. Without the assistance of these actors, such policies may have needed to be imposed rather than, as was the case in England, enthusiastically adopted.

Institutionalist analysis of these episodes of policy change needed to take account of historical, rational choice, organisational and discursive institutionalist explanations and would not have been complete without applying analysis based on all four logics of human action, namely learning from history, rational choice of the best option, compliance with rules and norms and the process of discourse about change itself.

Kingdon’s MSF was found to provide an excellent organising tool for diagnostic and prescriptive enquiry but was found to require a stronger focus upon the importance of institutional factors including ownership and governance arrangements, rational choice explanations for how actors’ incentives operate individually and collectively and historical policy-making antecedents as factors in policy change and variation. The MSF was also found to need a longitudinal focus to enable the full extent of the process of non-incremental change to be understood over time. These findings have much in common with Bakir’s findings that the MSF lacks key historical and structural components and ignores interactive processes in social and institutional contexts (Bakir, 2013). Without taking these additional factors into account, the ability of the MSF to display the full complexity of these two processes and to attain greater relevance for policy-making in Westminster jurisdictions was found to be limited.

Conclusion

The case studies show that national-level non-incremental health policy change can be achieved without conditions of ambiguity or manipulation by exogenous policy entrepreneurs, counter to the predictions of Kingdon’s MSF. A new analytical framework to find drivers of policy change and variation was used. Based on a most similar systems case study approach, the framework assessed each case study against the elements of the MSF then against five other explanatory approaches to policy change and variation, compared the results between case studies to find the points of difference and assessed these to establish whether there could be attribution of cause and effect.

While the structural and agency features were similar and vital to the success of the policy-making process in both countries, differing institutional drivers explained the policy change and variation observed. Historical institutionalist approaches were particularly predictive in these two case studies, as both demonstrate a strong influence of past struggles and bargaining on the respective policy outcomes. Organisational institutionalist concepts illuminate the challenges of pay-for-performance policy to the long-standing rules of appropriateness within the general practice profession, of professional autonomy and self-regulation, which were underestimated by NZ policy-makers but not in England. Rational choice institutionalist approaches convincingly explain the powerful logics of consequentiality which were seen to influence decision-making by general practitioners individually and collectively, especially in England. There, general practitioners seized the opportunity to enhance their individual expected utility through substantial increases in income and their collective expected utility (in preserving a strong negotiated framework for the continued coexistence between the profession and the state). In NZ, general practitioners did not do so to the same extent, leaving substantial performance funds unspent. Discursive institutional theory has been helpful in highlighting the influence of the distinctive differences in communication processes in these two policy-making processes in the two countries: bargaining amongst general practitioners which engaged the profession in England and broad-based primary care consultation which did not in NZ. A further important influence was the coordinative discourse used by particular actors at the centre of the processes of policy change.

Failure to use all four approaches, and to acknowledge the importance of agency, would have resulted in a partial explanation of these two complex processes of change and a risk that important learnings for future policy-making would be overlooked. These lessons include recognition that states may need to seek changes in governance and ownership arrangements within general practice to gain effective stewardship of their primary health care system and that effective policy-making within general practice may require mechanisms for collective action by general practitioners and the nurturing of positive working relationships between state funders and general practitioners. Future research might, in addition to the structural context for policy-making explored here, consider broader ideational and material contexts in which policies, institutions and entrepreneurs are embedded (Bakir and Jarvis ref from this Special Issue to be inserted in text and in Reference list).

Notes on contributors

Verna Smith is Programme Director of Masters Programmes and Senior Lecturer. The author’s key Interests are Multiple Streams Framework, Comparative policymaking, and Health Policy.

Jackie Cumming is Professor of Health Policy and Management Director, Health Services Research Centre Associate Dean, Research Faculty of Health Victoria University of Wellington. The author’s key interests are health policy, health system performance, access, and inequities.

Disclosure statement

No potential conflict of interest was reported by the authors.