The authors wish to thank the interviewees that provided their time and insights to support the research.

Introduction

1Policies driving major, complex health system reforms often have broad, ‘transformational’ aims involving new constellations of organisational arrangements and service provision configurations across scales and contexts. Such ‘large-system transformations’ (Best et al., 2012) are not generally characterised by a top-down engineering approach to meet policy aims. Instead, they involve a staged implementation process and seek broad engagement of system stakeholders to foster shifts in behaviour and organisational change. The dynamic and varied contexts of implementation, combined with the iterative and evolving implementation of such policies over time, make summative assessment challenging (Gray & Shaw, 2018) and presents challenges to traditional policy assessment approaches that aim to ascertain success through linear causal models and quantitative outcome measurement (Greenhalgh & Papoutsi, 2018; Pawson & Tilley, 1997). While attempts, rooted in a positivist position, to quantitatively assess the success of such reforms have been undertaken (Fatimah et al., 2023), the breadth of potential outcomes makes selecting appropriate indicators challenging and the potential for interconnected effects between indicators of success is difficult to disentangle in a meaningful way.

2Attempts to understand and grapple with the implications of this complexity of policy enactment and assessment have been made. This includes recognition that ‘layers’ of institutionally-mediated responses to policy can build up over time in variable ways within and across contexts (Mahoney & Thelen, 2009), which can result in unpredictable effects. Jones’ (2017) work on ‘sedimentation’, for example, highlights how layers of policy can build up over time, involving the need for local actors to navigate and resolve policy in ways that account for locally relevant effects of policy history. This has the potential to create nested, co-existing assessment frameworks where success itself is historically and geographically contingent, shaped by local policy legacies as well as explicit policy aims from the centre. Furthermore, in health policy particularly, such contingent assessments connect to deeper areas of normative contestation around equity, access, and individual and societal responsibilities (Stone, 2012; Klein, 2013). Generally, however, assessments of policy impact have adopted a relatively narrow temporal focus to support assessments of success and have tended to overlook the temporal dimension in trying to understand the development, implementation and implications of policies (Bressers et al., 2013). The consequence of this may be to underappreciate the role of previous policies and contextual factors in shaping the conditions and content of a policy of interest and, therefore, understanding how and why a policy is successful or not. An approach that can integrate a nuanced understanding of success in complex, evolving policies with a sensitivity to temporality, and the way previous policies and related history feed into shaping their detail and realisation, may offer a valuable way forward for policy assessment.

3In this paper, we explore integrating conceptual tools to support this by combining McConnell’s (2010a; Marsh & McConnell, 2010; McConnell et al., 2020) policy success framework with policy feedback theory (Béland et al., 2022; Mettler & Sorelle, 2018; Pierson, 1993). This combination offers the potential of policy assessment through a broader, more dynamic, temporal lens that provides explanatory power for understanding policy success and political change. We explore the value of combining a ‘shorter’ temporal focus on policy assessment with a ‘longer’ sweep of policy history and use dimensions of policy success to consider different types or orders of feedback. We seek to understand how policy feedback effects reshape the political environment in ways that affect future perceptions of policy success and failure. This is, to our knowledge, a novel systematic combination.

4To explore the value of this approach, we focus on a policy case study – the Health and Care Act 2022 in the English NHS – while also considering the implications of previous policy initiatives including those leading to the Health and Social Care Act 2012 and subsequent developments. The 2022 Act is notable because it represents a major shift from the market/competition logic of the 2012 Act in favour of collaboration and integration. This provides an ideal case for exploring how policy feedback shapes policy design and mediates the potential for success across process, programme, and politics dimensions. We highlight how feedback processes and competing feedback loops operate both within and across different dimensions of policy success. This creates both stability and change pressures that reveal some of the underlying dynamics that warrant consideration when evaluating complex, recursive policy processes. This highlights some of the particular challenges in assessing policy success and failure in complex sectors and demonstrates the importance and value of temporally and contextually sensitive approaches to policy assessment.

Integrating theoretical tools: Policy success framework and policy feedback theory

5McConnell’s (2010a; Marsh & McConnell, 2010; McConnell et al., 2020) policy success framework is a multi-dimensional framework intended to address some of the reductive tendencies in policy assessment where judgements regarding success are treated in a binary way. In the framework, success is understood as a spectrum within the dimensions of process, programme, and politics. Within each, success is considered broadly in terms of the extent to which a policy achieves “…the goals that proponents set out to achieve and attracts no criticism of any significance and/or support is virtually universal” (McConnell, 2010b, p. 351). Process success relates to the development of the policy and the extent to which it endures legislative or other routes to realisation. It also encompasses the creation and/or management of policy coalitions and stakeholder groups. Programme success depends on the fulfilment of the aims and objectives of the policy and how effective it proves in addressing relevant challenges. Success can also be graded between the poles of success and failure in each dimension: resilient, conflicted, and precarious. This approach highlights, and provides a means to examine, the multi-faceted and contextually embedded nature of policy success. It has been applied widely and used to assess a broad range of policies and sectors including housing (McConnell et al., 2020), environment and transportation (Callahan, 2019), and health services (Hammond et al., 2022; Checkland et al., 2023), amongst others.

6In a slight modification to McConnell’s (2010b) approach, in this paper we respond to Newman’s (2014) call for a more nuanced assessment of success, which emphasises that other perspectives beyond those of policy architects must also be considered. We therefore broaden questions of process and programme success to incorporate success as seen by actors other than those responsible for the policy. In McConnell’s original formulation, political success was seen as representing party political benefits. However, Checkland et al. (2023) argue for a broader notion of politics, suggesting that, because politics is fundamentally about the distribution of power and resources, assessments of political success should encompass: “how policies influence the distribution of power or resources in a political system alongside party political advantage” (Checkland et al., 2023, p. 467). This brings into focus NGOs, Arm’s Length Bodies, and other non-party political actors who nevertheless may accumulate power. When exploring and assessing policy involving multi-level system change, and a shifting emphasis from design to implementation, the number of actors with a legitimate stake in shaping policy increases and thus a pluralised notion of success becomes more important. Furthermore, following Bovens (2010), our analysis rests on a recognition that programmatic success can be assessed through both process and outcome lenses, i.e., examining whether a policy achieves substantive outcomes and whether the processes through which programmes are designed and delivered are appropriate and effective. This is relevant particularly to policies focused on governance and system change, where programmatic objectives may concern how decisions are made and services coordinated, not just end outcomes.

7Attention to temporality in policy analysis is well established, including work on path dependence (Pierson, 2000), policy cycles (Howlett et al., 2009), and temporal sequencing in historical institutional analysis (Thelen, 1999). More recently, efforts have been made to address a lack of temporal sensitivity in McConnel’s (2010b) framework (Luetjens et al., 2019; McConnell et al., 2020; ’t Hart, 2019). This includes work by Compton and t’Hart (2019) arguing that a public policy can only be considered successful when its desirable social outcomes are sustained over time and demonstrate resilience as circumstances evolve, and by McConnell et al. (2020), which explicitly incorporates time as a fourth dimension of success. Checkland et al. (2023) adapted the framework further to consider broader policy time horizons beyond specific pilot programmes and also noted the importance of being explicit about how the question of “success as defined when?” is addressed in a given policy assessment. However, attempts to integrate considerations of temporality into assessments of policy success have been confined primarily to assessing individual policies over time, as outlined in Table 1. They do not systematically explore how previous policy waves and related context mediate and shape the dynamics of policy development, implementation, and perceptions of success within the dimensions of process, programme, and politics. We address this gap in the current paper by calling on policy feedback theory.

Table 1. Different approaches to engaging with temporality in policy success literature

Temporal dimension	Luetjens et al. (2019)	McConnell et al. (2020)	Compton & ’t Hart (2019)	Checkland et al. (2023)
Integration approach	Sustainability as one of four success criteria alongside programmatic, innovation, political	Temporal realm as separate fourth dimension in assessment framework	Endurance as fourth dimension requiring "multi-shot sequence" assessment	Three-stage temporal structure: process, outcomes, longer-term effects
Scope	"Long-standing reforms" with "sustained public acceptance and bi-partisan political support" over time	How policies "differentially benefit a variety of stakeholders" at different temporal moments	How assessments change over time: "what you see often depends upon when you look"	Temporally bounded pilots transitioning to "enduring organisational arrangements"
Engagement method	Case selection based on demonstrated "genesis and evolution" over extended periods	Framework assesses success across temporal realm to avoid "rushing to judgment"	Continuous reassessment: "not through a one-off snapshot but as a multi-shot sequence"	Separating "outcomes of particular pilots from assessment of the longer-term effect of the pilots in influencing policy"
Complexity	Cases maintaining acceptance across multiple governments over decades	Temporal considerations help avoid premature success claims	Short-term and long-term effects may in some cases be at odds	"Temporal dimension of success implicit in moves from temporary pilots to enduring organisational arrangements"

Source: the Authors

8Policy feedback theory has roots in the emergence of Historical Institutionalism in the 1980s (Pierson, 2011) and efforts to ‘bring the state back in’ to political analysis, recognising the significance of states as autonomous actors that shape politics through institutional and administrative processes (Evans et al., 1985). The fundamental understanding of policy feedback is that policies are not just effects but also causes that “…restructure the political and social environment that produced them” (Pierson, 1992, p. 359). This includes institutional effects, driving new or reshaping existing political constituencies or groups, cognitive and interpretive frames, and resource distribution and access (Mettler & Sorelle, 2018). There are a number of different strands of policy feedback literature, including work exploring the effects of policy feedback on interest group mobilisation (Patashnik & Zelizer, 2013), political participation of citizens and policy attitudes (Campbell, 2012; Mettler & Soss, 2004), and institutional path dependence (Hacker, 2002) - see Béland et al. (2022) for a review. These strands treat feedback itself in different ways; in historical institutionalism, positive feedback refers to the amplification of any institutional trajectory, while in punctuated equilibrium theory, negative feedback refers to equilibrium-maintaining dynamics. In this paper, we focus on feedback as a driver of political and institutional change. Work in this vein has primarily considered how ‘positive’ or ‘self-reinforcing’ feedback can create a recursive dynamic of reinforcement of a particular policy or policies related to it. This typically occurs due to a policy affecting the creation or growth of stakeholder groups and interests, resources, and political dynamics. Attention has broadened to ‘negative’ or ‘self-undermining’ feedback, whereby policy effects create pressure to modify or abandon policies. Some of the complexity of interplay between positive and negative feedback within the same policy sphere has been explored (Daugbjerg & Kay, 2020), but the focus in the literature has primarily been on using feedback to help explain policy endurance and change rather than considering it as a mediator of policy success.

9Positive and negative feedback can lead to a range of effects which are delineated in the literature, comprising: resource; interpretive; lock-in; and learning. Resource effects are where policies shape political action and public perceptions because of material resources they are associated with (such as money, services, goods) or other benefits or disbenefits. Resources might have a direct effect or incentivise certain actions of policy relevance (Béland et al., 2022). Interpretive effects involve information transmission shaping stakeholder perceptions in some manner (Pierson, 1993). These might be particularly prevalent in our current social media age (Hao, 2025). Lock-in effects generally describe policy durability reinforced by positive feedback, and the concept is related to path dependence in historical institutionalism. Learning effects reflect the knowledge and understanding of stakeholders developed through policy processes and the manner in which this can shape future policy (Bennett & Howlett, 1992; Heclo, 2024). We propose that each feedback type operates distinctly within the different dimensions of policy success. For example, interpretive effects might primarily shape political success by influencing narratives about policy achievements or failures, while learning can be expected to shape programmatic success through design improvements informed by previous implementation experiences. Resource effects can be expected to cut across all dimensions, enabling or constraining process legitimacy (via stakeholder capacity to participate), programmatic implementation (through available funding and infrastructure), and political coalition-building (through the distribution of benefits and costs). Lock-in effects can reinforce process patterns (established consultation mechanisms), programmatic approaches (embedded service models), and political alignments (entrenched interest group positions).

10Our integration of policy feedback with McConnel’s (2010b) success framework builds upon an understanding that policies are part of evolving temporal sequences rather than discrete. As such, feedback from previous policies mediates the institutional, political and interpretive context shaping subsequent ones (Pierson, 1993). Tracing feedback mechanisms across success dimensions supports examination of not just the extent to which policies succeed or fail, but the ways feedback shapes success trajectories over time. More specifically, policy feedback can be seen as one of the factors shaping both future policy making and perceptions of whether any policy is considered a success. Feedback represents the stories which are told about policies, sometimes underpinned by concrete evidence of impacts, but commonly representing accumulating shared (or contested) perceptions, deriving as much from prior assumptions and political ideologies and even ‘group think’ as from robust evidence. Feedback is thus one of the strands in an ongoing temporal flux – in Kingdon’s (2003) terms, an important element in the policy ‘stream’ which may lead to the opening of ‘windows’. As such, an assessment of how and why a policy is seen as a ‘success’ across McConnell et al.’s (2020) various dimensions requires, amongst other things, an understanding of the stories that were and are being told about this and other policies, both in the past as the policy was developed and implemented (thereby influencing what happened) and at the point at which judgement of success is made. These stories influence what is deemed possible, what is seen as desirable, and what needs to be avoided. The stories which arise out of any judgement of ‘success’ will themselves become the substrate out of which future policy is made. Paying attention to policy feedback thus has the potential to broaden and deepen our understanding of policy impacts through time. Importantly, considering policy feedback alongside dimensions of success helps us to move away from discrete episodes of policy assessment – asking what happened to each particular policy at a moment in time – to a more dynamic and unfolding picture of interwoven policies, with feedback influencing not only each policy as it is implemented and judged but also other related policies in adjacent fields.

11Such a dynamic approach to policy assessment does bring with it some challenges. Opening policy assessment to consider these wider influences could lead to an infinite iteration, in which chains of feedback and impact through time cannot be straightforwardly separated and considered and, therefore, intentional and logical ‘bounding’ of scope is required. Nevertheless, we believe that there may be value in bringing together these two areas of policy theory to support a shift in evaluative focus: moving from attention to whether or not a policy is successful to exploring how that policy is performing over time and examining what feedback mechanisms are influencing that performance (and how these may come to shape the success of the policy in the future). This paper therefore addresses the question: How have policy feedback mechanisms shaped the trajectory of policy success across process, programmatic, and political dimensions in English health and care integration policy?

Methods

Case selection strategy and background

12The focus of this paper is the 2022 Health and Care Act in the English NHS but with sensitivity to the feedback effects associated particularly with a previous large-scale reform – the Health and Social Care Act 2012 – and the policy developments that followed. The 2022 Act was selected because of its relevance to the assessment challenges identified previously, and because it represents a point of culmination for a diffuse stream of policy around inter-organisational integration and collaboration, which allows us to explore feedback effects over time. The policy is a politically salient large-scale system transformation affecting multiple organisational levels and stakeholder groups and thus provides analytical utility when considering how feedback may shape multidimensional success.

13The Health and Social Care Act 2012 was a significant reorganisation introduced by the Conservative-Liberal Democrat Coalition Government. Largely unanticipated in advance (and therefore representing an unusual piece of legislation which preceded rather than followed policy discussion and development), it introduced sweeping reforms designed to cement competition as an animating principle and to give clinicians a prominent role in service commissioning (Timmins, 2012; Checkland et al., 2018). Existing local and regional statutory bodies were abolished and replaced with smaller GP (i.e. family doctor) -led Clinical Commissioning Groups. These groups were charged with commissioning services for a defined geographical population from a market of competing providers. A new arm’s-length body, NHS England, was introduced to oversee the new system, with responsibility for commissioning some highly specialised services as well as resource distribution and performance management, operating under an annually refreshed mandate from the Secretary of State for Health. Central government thereby relinquished operational control of the health service, with the expressed intention “to take politics out of the day-to-day management of the NHS” (Timmins, 2012, p. 25).

14A number of features of the legislation were intended to enhance market competition and patient choice, including: bringing the NHS within the purview of European competition law (Osipovič et al., 2020); strengthening the regulation of competition, with an existing arm’s length body (Monitor) given specific responsibility for enforcement; and the introduction of ‘Any Qualified Provider’, which was a measure intended to increase patient choice over where they would receive care from any providers meeting certain standards and costs in order that money could ‘follow the patient.’ Other key elements of the 2012 Act – particularly decentralisation/localism and clinical leadership – were seen as advantageous because it was assumed that GPs would improve service quality and efficiency by making clinically informed commissioning decisions (Checkland et al., 2018). Moreover, such decisions would be more attuned to local needs because the approximately 211 Clinical Commissioning Groups occupied smaller footprints than the 152 Primary Care Trusts that preceded them as commissioners of the majority of primary, secondary, and community services.

15In the decade following the 2012 Act, UK and international policy began to shift away from competition towards a recognition that providing care to an aging population with multiple health conditions required a more integrated approach (Cash-Gibson and Rosenmoller, 2014). Influential UK policy think tanks such as the King’s Fund started to push integration as a policy solution (Addicott, 2014) and, in 2014, NHS England took advantage of its operational independence to publish a manifesto for greater service integration: the Five Year Forward View (NHS England et al., 2014). The Forward View emphasised the need for local health communities to be supported and empowered to work more effectively together. A number of models were proposed, alongside a national piloting ‘Vanguard’ programme which would support the bottom-up refinement and development of these ‘new care models’ which could then be more widely adopted (Checkland et al., 2019). Crucially, selected pilot sites were provided with significant funding and operational leeway to develop their approach to more integrated care. Alongside the pilots, an extensive evaluation programme was established (Operational Research and Evaluation Unit, NHS England, 2016), early findings from which were quickly presented as demonstrating their ‘success’ (Berwick, 2016). Soon after the initiation of the Vanguard programme, and before its effects could be fully assessed, NHS England published its Long-Term Plan (NHS, 2019a). In this document the Vanguard pilots were presented as an unequivocal success, in part driven by a political need to convince the UK Treasury that additional investment in the NHS would not be wasted. Importantly, the extent to which the Vanguard programme was successful in programme terms (i.e. achieving its ostensible goals) was less important than its political effect in reinforcing the perception that local collaborative action within a loose national framework had been a successful approach to reform (Checkland et al., 2023). The COVID-19 pandemic provided the conditions for further local inter-organisational collaboration, as performance regimes were loosened and financial flows streamlined (Sanderson et al., 2022). The result of all this was that by 2020 a new policy narrative had taken hold in England, whereby local collaborative activity between providers, with collective responsibility for achieving financial balance, was seen as a necessary step in putting the NHS on a more sustainable footing (NHS England, 2020). It was against this background that the 2022 Health and Care Act was developed.

16The Health and Care Act 2022 saw the legal abolition of Clinical Commissioning Groups and their replacement with statutory Integrated Care Systems (ICSs; 42 in total) with more diverse organisational composition, and an emphasis on shared consensual decision making between local health and care organisations. In practice, the transition from Clinical Commissioning Groups to ICSs had been underway for several years (with ICSs, and their predecessors Sustainability and Transformation Partnerships/STPs, existing as non-statutory informal network arrangements as early as 2016), illustrated by widespread mergers between Clinical Commissioning Groups to create larger footprints. Their explicit aims are to: “Improve outcomes in population health and healthcare; Tackle inequalities in outcomes, experience and access; Enhance productivity and value for money; Help the NHS support broader social and economic development” (NHS England, n/d). The Act signalled a marked move away from provider competition in favour of collaboration, and a reduced emphasis on markets and market mechanisms. There was also a rhetorical shift of emphasis to a population health approach alongside a renewed commitment to prevention. The formal provisions of the Act established Integrated Care Boards (ICBs) as the new statutory body, with each ICS given freedom to determine for themselves aspects of their ICB composition, although there was an expectation that senior executives from local providers would be represented.

17In each ICS, an accompanying Integrated Care Partnership (ICP) was required to be established, bringing together NHS representatives with those from Local Authorities and the Voluntary Sector. The ICPs’ role was to assess local needs and develop strategic plans, with ICBs charged with their operationalisation. The policy emphasised the importance of providing ICSs flexibility to organise governance and operations locally, and the Act was permissive, with membership, roles, remits and accountabilities of the different bodies left up to Systems to decide beyond minimal stipulations for ICBs (NHS England et al., 2021). However, it was suggested that Systems should operate under a principle of subsidiarity, with an expectation that ICBs will delegate resources and decision-making powers to lower geographical levels known as ‘places’, although this was not mandated. The 2022 Act further sought to support system working by enabling NHS England to modify licence conditions to enable co-operation and by introducing the ‘Triple Aim’ duty (a duty on NHS organisations to consider the effects of their decisions on the better health and wellbeing of everyone, the quality of care for all patients, and the sustainable use of NHS resources). All of this occurred in a context of marked financial pressure and a requirement that ICBs control spending and find ways to realise savings.

Data sourcing

18Fifteen semi-structured interviews were conducted with national level policy makers from Government departments and arm’s length bodies, as well as with individuals from organisations representing groups of health and care stakeholders (May-October 2024). This was a work stream in a broader mixed-methods post-implementation review of the Health and Care Act 2022 study. The aim of this study is to develop evidence regarding collaboration within and between Integrated Care Systems in the English NHS in relation to the provisions of the 2022 Act. Interviews were conducted by [Removed for anonymity]. These were audio recorded and professionally transcribed. Limitations of this methodological approach, including sample size and reliance on self-reported perspectives, are considered in the Discussion.

19Ethical approval for the study was obtained by [Removed for anonymity]

Data analysis

20Transcripts were pseudonymised and analysed using NVivo 14. A coding framework was developed by [Removed for anonymity] to address the overarching project’s aim of understanding inter-organisational collaborative health and care activity in the context of the 2022 Act (see Appendix 1 for an extract of the coding framework). At the start of the analysis process, several transcripts were double coded within the team and the coding frame developed through subsequent discussion. The role of history in shaping the policy’s form, stakeholder expectations, and understandings of successes and challenges relating to the policy was prominent in this analysis, which led to engagement with the relevant literature explored above. This provided the basis for the ex-ante theorisation provided in the previous section, and a secondary analysis by the authors – focusing on dimensions of policy success and the identification of policy feedback relevant phenomena – was then conducted. Interview extracts presented below contain interviewee pseudonyms in square brackets with the month and year of the interview.

21A collection of policy documents and legislation were also subject to analysis to characterise ‘official’ policy aims at a given time and to understand how these evolved. This was used as a point of reference in relation to the interviewee narratives. The documents analysed were: Health and Social Care Act (2012; and associated guidance); NHS Five Year Forward View (NHS England et al., 2014); New Care Models: Vanguards – developing a blueprint for the future of NHS services (NHS, 2016); NHS Long Term Plan (NHS, 2019a); Health and Care Act (2022; and associated guidance).

Operationalising policy feedback mechanisms

22Here we operationalise how policy feedback effects are identified empirically in our analysis, drawing primarily on interview data with policymakers and system stakeholders alongside policy documents. As noted, we conceptualise feedback as a driver of political and institutional change, with positive or ‘self-reinforcing’ feedback and negative or ‘self-undermining’ feedback creating a recursive dynamic of reinforcement or undermining of a particular policy or policies.

23Interpretive effects are traced through how stakeholders frame and perceive previous policies, including problem definitions, narratives of success or failure, and perceptions of what is possible or desirable. Resource effects are identified through changes in material resources, organisational capacity, and accountability mechanisms stemming from previous policies. Learning effects are evidenced through explicit incorporation of lessons from previous policy experiences into subsequent policy design. Lock-in effects are demonstrated through the persistence of institutional arrangements, accountability structures, and coalition commitments despite changed circumstances.

Analysis

24In this section, McConnell’s dimensions of policy success are used to structure the analysis, and policy feedback effects are considered within each.

Process success and feedback

25The process success dimension relates to the formulation of the policy and its endurance through this towards implementation. It includes attention to policy coalitions and stakeholder groups and how these have been created or managed in relation to this. This aspect is particularly pertinent for the 2022 Act because a coalition of support around it was generated through policy design and communication that explicitly contrasted it with the 2012 Act’s approach. Below we consider how negative feedback relating to the framing of the 2012 Act shaped process dimension activity, and the role of positive feedback in fostering a coalition in favour of the 2022 Act’s approach.

26Interviewees reported that the 2012 Act had generated mistrust of Government policymaking amongst system stakeholders. This was traced back to the aims and provisions in the 2012 Act to encourage and require competition between service providers in the commissioning process, as well as to the structures and requirements on individuals working in clinical commissioning organisations, which were perceived as restrictive and burdensome. The 2012 Act framed NHS inefficiency and lack of responsiveness as problems requiring market solutions, but in practice local commissioners found that it was often necessary to circumvent competitive processes in order to enable logical local collaborative arrangements (Osipovič et al., 2017). Consequently, those orchestrating services instituted workarounds that were inefficient and demanding (Checkland et al., 2018).

27This negative interpretive feedback from the 2012 Act shaped the 2022 Act in multiple ways. The 2022 Act and related guidance identified fragmentation and competition — and by extension ‘commissioning’— themselves as problems requiring a policy solution, which was in the form of integration and collaboration. Thus, the 2012 Act was itself part of ‘the problem’ being solved. In addition, the new Act involved a swing away from the structured and prescribed approach that the 2012 Act took to how Clinical Commissioning Groups should be constituted and operate towards an approach involving a significant degree of permissiveness and flexibility around how ICSs could organise themselves, particularly regarding governance arrangements. Indeed, so strong were the negative perceptions of the 2012 Act, policy makers reportedly perceived that emphasising the difference in approach between the two Acts in this way was necessary to establish the requisite trust from stakeholders and generate sufficient ‘buy in’ to secure successful transition to the programme phase.

28Positive interpretive feedback associated with increased permissiveness was reportedly in effect amongst policymakers during the development of the Act, as they tried to outcompete each other in the extent of permissiveness in the provisions being put forward. This suggests feedback whereby permissiveness became a policy characteristic that extended beyond a practical mechanism to an indicator of enlightened policy design:

well if this isn't the route to value, if it is perfectly alright for providers and commissioners, or providers and planners as we might call them now, to sort of get together and work in joint spaces, then why can't we just take down all the walls, and kind of be very, very, very permissive about all of this, and so we almost became competitive with each other about how permissive we could be [Q6N2, Aug24].

29Policymaker interviewees talked about the shift from mandating structures and dynamics within organisations and systems to the 2022 Act as a means of ‘permission-giving’ through a strategic process approach. One explained:

The law can’t make people behave, I mean it can a bit, but you can’t just decree ‘thou shall collaborate’ from the centre and people will work together, it doesn’t work like that. I think what it did was give the people who wanted to do more collaboration, the people who wanted to start thinking of systems, ammunition in that argument and was a very clear marker that the way the government was viewing the health service had changed. That it was no longer a competition-y, market-y place and it was in a much more collaborative place. [SOPL, Sept24]

30This highlights an interplay of multiple feedback mechanisms. The 2022 Act, as espoused, was intended to deliver resource effects to system stakeholders in favour of collaborative working by giving them ‘ammunition’ in the form of legitimising political support for their preferred manner of service orchestration. This ‘clear marker’ of how the Government viewed the health service also constitutes an interpretive effect supporting a coalition defined in part by preferences ‘to do more collaboration.’

31‘Permission-giving’ also involved delivering reassurance to system stakeholders by legitimating actions that were avoided due to perceived rather than actual constraints: "very often people think there are technical and legal barriers in the way to do things, that actually there aren't when it comes down to it" [HD90, Aug24]. This illustrates interpretive feedback loops in action, whereby perceived constraints limit potential actions that require interpretive process meditation that extends beyond just practical or technical legislative change. The very permissiveness of the new Act was in part designed to support this new interpretation.

32The desirability of the shift away from competition to collaboration was also seen as being driven by the response to the COVID-19 pandemic. This included suspension of the usual activity-based payments system in favour of block contract payments, and the development of joint system responses to local service provision with improved inter-organisational communication. Although this was partially a result of the suspension of certain usual rules and requirements, interviewees reported that it nonetheless created certain learning effects that fed back into the development of the 2022 Act and the desirability of the collaborative mode. These learning effects were layered onto those associated with the claimed success of Vanguards, as well as a broader direction of travel running through the Five Year Forward View and Long-Term Plan.

33This account highlights how feedback from the 2012 Act’s implementation, and the subsequent narrative around that legislation, did not just create interpretive policy effects that shaped the characteristics of the 2022 Act, but also mediated the broader process of policy development. This suggests a recursive dynamic whereby the policy development process was also a means of policy feedback management. Furthermore, it demonstrates how policy feedback is layered across different temporal scales simultaneously, with negative interpretive feedback associated with the 2012 Act interacting with positive learning feedback from the COVID-19 pandemic, and the positive resource effects around the legitimisation of collaboration supporting coalition development and preservation. The successful passing of the 2022 Act and the development of a supportive coalition, mediated particularly by interpretive effects, suggest resilient process success overall.

Programme success and feedback

34Programme success is established based on how well any policy has achieved its explicit objectives. For the 2022 Act, programme outcomes are more mixed when assessed now. In some respects, positive feedback is supporting increased collaboration, however, negative feedback due to accountability measures in tension with policy aims is undermining this. Programme success is also inevitably a more open question due to the unresolved nature of the development of policy around the 2022 Act at this stage, and because the broad policy aims – which were helpful in securing broad ‘buy in’ and perceived as self-evidently desirable aspirations – are not straightforward to assess because of the challenge of disentangling other influential factors mediating success and failure in diverse local contexts.

35With these challenges noted, the 2022 Act has demonstrated programmatic success in the sense that it formalised and legitimised certain collaborative ways of working with systems – removing legislative barriers and the need for workarounds and reflecting positive learning effects by supporting policy that responds to the need for these informal approaches. Interviewees reported, however, that policy aims around devolution of organisational functions and the principle of subsidiarity were perceived as not being met in practice. A significant part of the explanation for this relates to the issue of accountability and organisational sovereignty. While guidance suggests that the principle of subsidiarity is expected to be embedded in ICB arrangements, with responsibility for the planning and delivery of integrated services delegated to geographical places (NHS, 2019b), legislation does not address the issue of delegation or devolution of ICB functions to system forums (House of Commons, 2021). The Act’s ‘duty to collaborate’ provisions attempted to share responsibilities for population health improvement between health and care organisations, and a new duty for NHS providers to each have regard to the system financial objectives attempted to create collective accountability for system financial performance. However, providers remain separate organisations with their own accountabilities and statutory responsibilities which they must hold in regard when agreeing collective system plans. For example, NHS Trusts and Foundation Trusts (FTs) have legal duties to provide safe care and treatment and direct accountability to the centre (currently Department of Health and Social Care via NHS England) for their financial performance. The centre has limited levers to affect ICS behaviour, particularly in a context of constrained funding, and therefore there is a tendency to require regular financial reporting from the ICB to demonstrate ‘grip’ to Government ministers. One interviewee explained the behavioural impetus for this:

There is an emotional nervousness of decision-makers in the centre, both NHS England on one side but also ministers […]. If they can measure something, if they can see it all on the data, that’s the same as accountability and that makes them feel reassured. I think if you asked ICSs whether they were being held accountable for that, whether it was the right thing to be held accountable for and whether it made them better, I think you would get a unanimous set of noes, so I think there is an anxiety there. [SOPL, Sep24]

36This ‘emotional nervousness’ of ceding control from the centre creates negative feedback that reduces the willingness of ICB leaders to delegate control over budgets to more local ‘places’ within the systems and therefore undermines the espoused principles of subsidiarity:

So, I don't think we're seeing a lot of, if any, financial delegation to place. I think in some ICSs, they recognise that the place is a strength and they've got place-based leads and they've got Place-based Partnerships and... to varying degrees, those voices are recognised and listened to and trusted. But I don't think we've seen real concrete delegation to those place leads [IG30, May24].

You know, so that...that idea of subsidiarity is the...it's a...it's a goal to aim for. But I think very few places are actually achieving it because they...their attention is necessarily, kind of, constrained on how to control costs from the top-down [W570, Jul24].

37This highlights how resource effects and accountability arrangements via existing performance management approaches create negative feedback dynamics that undermine policymakers’ objectives around subsidiarity and delegation. One interviewee articulated the central problem: "what's measured, gets done" [I310, Apr24]. There is the prospect that this generates a negative feedback loop that progressively undermines the policy’s core aims, as measurement and financial monitoring (interpretive effects) associated with financial pressures (resource effects), reinforce centralisation of control whilst simultaneously undermining the conditions for successful inter-organisational collaboration and delegation.

38This programme dimension analysis highlights that positive feedback through learning effects may be insufficient to deliver success in complex health system reforming policy when negative feedback associated with structural accountability dynamics is operating in tension. The broad policy aims associated with the 2022 Act, which helped to build a coalition around implementation and included an emphasis on the value and importance of subsidiarity and local decision making, may undermine ongoing support when stakeholder expectations are not met because of conflicting feedback. There is an important temporal consideration here, too, as short-term accountability pressures undermine the longer-horizon collaboration building processes associated with delegated decision making. Overall, with the previous caveats acknowledged, this suggests a provisional ‘conflicted’ assessment of programme success.

Politics success and feedback

39Political success refers to how the policy affected the political standing of those responsible for it. This is assessed here in relation to the policymaker interviewees associated with the 2022 Act and related policy guidance. A strong claim to success in this dimension relates to the strategic narrative associated with the 2022 Act that it was part of a policy programme that was built on reflecting upon past failures, learning from these, and being ideologically pragmatic. This suggests positive interpretive feedback whereby the policy issues associated with the narrative around the 2012 Act were employed as a source of legitimacy for policy makers.

40Generating a shared understanding of the problem (fragmentation and barriers resulting from pursuit of competition) and the prescription (collaboration, integration, flexibility and permissiveness) was politically helpful for policy makers in building a coalition around the policy to drive ‘buy in’ and implementation. This generated resource effects for policymakers in the form of system stakeholder support whilst also driving lock-in effects for those parts of the coalition formed around the belief that increased collaboration was the desirable policy solution. Significantly, by framing collaboration and integration as emblematic of policy competence and casting competition (and by extension commissioning, conceptually at least) as a regressive approach linked with policy failure, an interpretive feedback loop is created whereby policymakers benefit from credit associated with the former and avoid blame associated with the latter.

41The policy emphasis on the desirability of permissiveness is important from a political success perspective. As a general approach, vesting significant responsibility for organising governance, roles, and remits in inter-organisational systems reflects a ‘bright idea’ (rather than a ‘blueprint’) approach to implementation (Harrison and Wood, 1999). This has the potential to politically insulate policymakers from criticism of ICS performance or issues where this is plausibly linked to locally determined arrangements. However, this permissive approach contained negative feedback potential, in that the resulting variability local arrangements may lead to increased variability in health and care outcomes. One interviewee recognised that there was a political limit to tolerance of variation of this kind, particularly if the impact on patients and populations became starkly contrasting [OD20_2May24]. This was seen as somewhat of a bind because reducing permissiveness and flexibility over local arrangements might constrain systems with established arrangements operating well, which would also come at political cost:

Do we legislate for that or do we put statutory guidance out for that or do we just let systems do what they want because some systems have been doing it for a number of years, some systems haven’t even got started and a patient’s experience of those two systems is going to be vastly different. So, variation is a huge problem, a huge problem. [I310, Apr24].

42This highlights how positive feedback associated with permissiveness, particularly arising from those systems with more established collaborative arrangements that do not wish to be ‘done to’ and disrupted, can operate in tension with negative feedback associated with variation between ICSs which may create pressure for greater structural standardisation and challenge political sustainability.

43This analysis suggests resilient success in the short term, but the dynamics are already in place for this to shift to conflicted or even precarious success, potentially rapidly. It highlights how interpretive feedback effects from this kind of policy approach can require ongoing maintenance and attention. Generating political capital and building a coalition for change by learning from previous perceived policy failures signals competence and establishes legitimacy. This approach then creates pressure for further adaptation in response to implementation issues to demonstrate ongoing learning and dynamism. This creates programmatic vulnerabilities over longer temporal windows because political success is rooted in the coalition’s dominant understanding of what is working and what requires attention and adaptation. This suggests that realising and sustaining political success depends upon policymaker skills in meta-governance (Jessop, 2004; Dommett and Flinders, 2015; Hammond et al., 2018), which in this context involves the management of feedback relationships not just within one success dimension but also between dimensions. This leads us to consider cross-dimensional feedback dynamics in more detail.

Cross-dimensional feedback dynamics

44The 2022 Act case study helps us understand that policy feedback is not just temporal, i.e. a past policy influencing a current policy, but dimensional, in the sense that feedback effects traverse success dimensions to shape policy success. This gives us a new depth and dimensionality to policy assessment that moves beyond a focus on whether a policy succeeds or fails, to consideration of how different types of success or failure mediate feedback dynamics that shape policy prospects. However, this does create further particular challenges for policy assessment. In this section, we explore this in more detail to support understanding of the contours of these challenges and thus how to engage with them more purposefully and effectively.

45The analysis above has revealed two broad cross-dimensional feedback loops in operation. The first is a ‘collaboration loop’, whereby: perceived programmatic policy success since the 2012 Act (relating particularly to the framing of the success of the Vanguards and the response to COVID-19) generates system stakeholder support for, and coalition building around, collaboration and integration; this delivers political success, partially as a result of taking a reflexive/learning policymaking approach, and also supports a permissive programmatic approach because key stakeholders expressed the desirability of this. These conditions are intended to remove barriers to collaboration and provide institutional support for collaborative activity at local levels.

46The second is an ‘accountability loop’: existing organisational accountabilities and performance management processes are unchanged; in a context of financial pressures, this leads to the centre exerting control over ICBs via reporting requirements to exert ‘grip’ and allay ‘emotional nervousness’; delegating control and responsibility to local levels within systems is undermined by this, which interferes with the collaboration intentions of the policy and thus affects system performance.

47These loops create a tension between stability (collaboration valued and politically supported) and change pressure (meaningful delegation undermined). One interviewee articulated this:

I think [...] most systems would argue that actually the reason you're delegating is because you can make change happen... but in an environment where the focus and the attention on operational performance and finance is so high... there's been quite a high degree of caution about moving forwards in that area [FIJ1, Sep24].

48These loops illustrate a number of feedback dynamics between dimensions. For instance, the collaboration loop highlights how generating stakeholder support and building a coalition in pursuit of process success created political capital (and success), which then supports a drive for permissive policy (programme). The accountability loop shows how accountability functions limits delegation (programme) which interferes with collaboration and created political pressure (political). A variety of cross-dimensional feedback effects can also be identified. For example, process success creates programmatic expectations from stakeholders in terms of subsidiarity in pursuit of the realising the broad overarching aims of ICSs as truly integrated inter-organisational systems. This is then undermined by the accountability regime. Programmatic issues at a national level, such as unacceptable variations in outcomes between systems, could then feed back into the politics dimension and create pressure for programme change which may involve increased standardisation.

49All of this points to deficiencies in assessment approaches that engage with success dimensions as independent variables. We have shown how feedback effects traverse success dimensions and how this generates evolving tensions and change pressures that reveal some of the mechanisms that underlie and mediate assessments of success and failure. We will consider this more fully in the Discussion.

Discussion

50Contemporary health and care system policies in England are complex. They often have ‘transformational’ aims involving new, iterating constellations of organisational arrangements and service provision configurations across scales and contexts with significantly variable local implementation. This creates challenges for making judgements regarding their success and failure, compounded by a lack of attention to the way that their design, implementation, and evolution is shaped by the political forces and coalitions relating to experiences of previous policies. Engaging more intentionally with this temporal dimension holds the promise of increasing explanatory and evaluative power.

51In this paper, we have used McConnell’s dimensions of policy success and ‘nested’ consideration of policy feedback into the framework, which provides a means of integrating temporal analysis and feedback dynamics into the dimensions themselves. This nested approach makes theoretical contributions to both policy success assessment and policy feedback theory. The analysis highlights that success dimensions can be understood as more than static evaluative containers, instead representing dynamic and inter-connected feedback production and adaptation systems, where feedback in one dimension shapes feedback in the others. The implication is that policy feedback not only operates temporally in the way it is traditionally understood – i.e. past policies shaping current policy – but dimensionally, too.

52Our analysis of the Health and Care Act 2022 addresses the research question: how have policy feedback mechanisms shaped the trajectory of policy success across process, programmatic, and political dimensions in English health and care integration policy? The analysis reveals how feedback shaped success trajectories through cross-dimensional dynamics. Interpretive feedback from the 2012 Act's perceived failure shaped coalition-building and policy design for the 2022 Act, while persistent lock-in effects in accountability structures constrained programmatic implementation despite policy rhetoric about delegation and flexibility. This created competing feedback loops operating simultaneously, which reinforced collaboration narratives while undermining programmatic aims. This demonstrates how feedback operates not just temporally (past shaping present) but dimensionally (success in one dimension generating tensions in others). Table 2 presents policy feedback effects in relation to policy success dimensions for the Health and Care Act 2022.

Table 2. Policy feedback effects and policy success dimensions in relation to the Health and Care Act 2022

		Policy feedback effects
Policy success dimensions		Interpretive	Resource	Learning	Lock-in
Process	+ve	Permissiveness valued; 2012 Act framed as problem to solve	Legitimacy and "ammunition" for collaborative working	COVID response and Vanguards validated collaboration	Coalition commitment sustained around collaboration
	-ve	2012 Act generated mistrust requiring explicit contrast	-	-	-
Programme	+ve	Competition reframed as fragmentation/problem	-	Workarounds from 2012 Act informed barrier removal	-
	-ve	"What's measured gets done" constrains subsidiarity	Financial pressures and reporting requirements undermine delegation to place level	Insufficient against accountability pressures	Organisational statutory responsibilities and central-local accountabilities persist
Politics	+ve	Reflexive policy making narrative signals competence; Credit for collaboration, blame avoidance for competition	Stakeholder support and coalition mobilisation	Learning approach builds legitimacy and demonstrates responsiveness	Coalition locked into collaboration
	-ve	Variation between ICS may erode narrative	Permissiveness creates variation risk	-	Requires ongoing narrative management

Source: the Authors

53We focus on a case study of the Health and Care Act 2022 in the English NHS but with a sensitivity to the consequences over time stemming particularly from the Health and Social Care Act 2012. We see how the 2022 Act was shaped by feedback resulting particularly from the 2012 Act and consequent policy. This forged a common political constituency across policy makers and a cohort of influential health and care system stakeholders, particularly service providers and commissioners, with a shared interpretive frame of what had caused the previous policy to ‘fail’ and therefore how the underlying philosophy, requirements, and provisions associated with the 2022 Act needed to represent an oppositional alternative. This analysis highlights how policy feedback mediated the development, design, and implementation of the 2022 Act. Countervailing forces within the policy were created whereby policy aims existed in tension with mechanisms and details of implementation. This highlights a fundamental mismatch between the philosophy and detail of the policy with the broader structures and forces shaping the implementation environment, and this has been shaped by positive and negative policy feedback.

54There are several specific policy assessment insights to be gleaned from this. Firstly, multiple feedback loops can be active simultaneously, potentially contributing to success in one dimension yet undermining prospects for success in another. Secondly, the central importance of temporality is clearly demonstrated, for instance, in the way short-term political success may create conditions that undermine longer-term programmatic success prospects. Thirdly, cross-dimensional spillover effects have the potential consequence of driving success in one dimension while simultaneously undermining success in another via the same set of feedback dynamics. This demonstrates the fundamental inter-connectivity of success dimensions. Considered in a broad sense, this case suggests that assessment approaches applied effectively in less complex policy environments may be insufficient for policy areas characterised by large-scale, evolving system reform, multi-level governance, and many policy ‘layers’ built up over time. As such, the conceptual and methodological tools for assessment may need to be refined further to support understanding of multiple, potentially conflicting feedback loops, and temporal effects in these policy spaces.

55The case study analysis reveals several policy process dynamics that are important to highlight. The design of the Health and Care Act 2022 was clearly shaped by perceptions of the implementation and consequences of the 2012 Act, particularly perceptions of restrictive competition-related requirements that necessitated practical workarounds, as well as a perception that the 2012 Act itself had been too prescriptive. A coalition around the 2022 Act was developed in part by contrasting the approach of the policy with the 2012 Act as a means to establish legitimacy and build trust. This came along with an emphasis on local permissiveness and flexibility in an attempt to counteract negative feedback associated with the legacy of the 2012 Act. This highlights how the development of coalitions around particular policy trajectories can become sources of feedback for future policy change and affect prospects of programmatic success. A particular risk here is that coalitions that form around broad aims can generate negative feedback when realities of implementation do not align with the expectations cultivated during policy development in the process dimension. Feedback from the coalitions that have evolved further as a result of the 2022 Act will undoubtedly shape future policy. Given all of this, assessments of policy success might usefully ‘look forward’, focussing beyond the extent to which policies meet their objectives, but also considering how the process of trying to realise these objectives will mediate coalitions that are both products and producers of feedback that will shape future success prospects. Policy makers may benefit from considering: how achieving success in one dimension may generate feedback that mediates the prospects of success in another; the varied and overlaid timeframes relevant to policy feedback processes that shape policy success prospects; how, on a more applied note, accountability structures and processes can undermine collaborative intentions, which may make a focus on alignment of performance frameworks a necessary success pre-condition.

56The contingency and vulnerability of political success as this connects to process approach is a notable finding. A key factor in the political success of the 2022 Act was the broader policy approach of demonstrating learning from perceived previous policy failures and using this to support coalition building. The consequence of this, however, is establishing expectations that an adaptive and reflexive approach will be sustained by policymakers, which may prove challenging over longer timeframes. The fundamental challenge it establishes is that political legitimacy becomes contingent on demonstrating ongoing learning through implementation. When implementation problems arise, the boundary between programmatic and political struggles is more porous. More broadly, this suggests that in complex policy domains, such as health and care system reform, the ability for policymakers to effectively meta-govern may be particularly crucial in order for them to manage feedback across dimensions in effective ways.

57The analysis presented also invites consideration of several challenges regarding assessments of policy success. Traditionally, assessment approaches assume a linear relationship between policy input and outputs and, where they adopt a multi-dimensional understanding of success, tend to treat these dimensions as discrete. The analysis presented above illuminates limitations in this understanding, both reinforcing and extending existing critiques of such assumptions. Firstly, assessing success is complicated further when positive and negative feedback are recognised as operating across dimensions. The feedback associated with resilient process success of the 2022 Act is fundamentally inter-connected and dynamically engaged with the conflicted programmatic success and (arguably) precarious political success in those dimensions. The nuanced, multi-dimensional dynamics may not make definitive declarations of overall policy success easier but they do offer enhanced explanatory power and more refined analysis. Secondly, the accumulation and interaction of feedback over time (re)shapes assessments of policy success in ways that reinforce the temporal contingency of such assessments. In other words, clearly answering the question “success as defined when?” is fundamental to the evaluative endeavour for any policy. Thirdly, when policies are intended to adapt and evolve based on feedback, then policymakers’ responsive abilities in this regard can be used to make claims about policy success, which reduces the scope to analytically ‘fix’ a policy in space and time to make success assessments (Hammond et al., 2022; Checkland et al., 2023).

Limitations

58Data in the paper reflects self-reported perspectives of policymakers, although these were situated in relation to the ‘official’ aims of policy via analysis of documents. Specifically, our assessment of policy success relies on a relatively small number of interview participants (n=15), which may not capture the breadth of stakeholder perspectives on success and failure. Although we referenced interview data against policy documents to situate perspectives in relation to explicit policy aims, alternative data sources (e.g. a more systematic analysis of stakeholder positions in consultation responses) or integration with other empirical research on English health policy implementation, particularly with a different time window, might yield different assessments of success, especially regarding programmatic outcomes. Our findings should therefore be understood as one perspective on policy success, shaped by our particular methodological approach and timing. The paper focuses on a specific policy case study, albeit utilising a broader temporal lens to support understanding of feedback, and therefore future research may usefully explore the utility of nesting policy feedback within success dimensions when applied to other policies and public policy sectors. It might be that the health sector is particularly prone to some of these dynamics because of its institutional complexity, professional autonomy, and political significance. The adapted framework should, however, be applicable to other policy contexts and doing so may elicit further insights and understanding. Furthermore, contrasting policies where cross-dimensional feedback dynamics are managed in a manner that supports assessment of success in multiple dimensions with those where they are not could be a valuable analytical approach.

Conclusion

59This analysis suggests that policy success associated with complex health system reforms (and potentially contemporary public policies more broadly) may be contingent on policy design that recognises and understands policy feedback dynamics and how these can shape success within and across dimensions. This suggests several practical lessons for policy makers: first, the policy design process should involve recognition that interpretive effects from previous policies shapes stakeholder receptivity and coalition-building potential; second, it is important to recognise that different types of feedback may operate in tension, e.g. negative resource or lock in effects may undermine positive learning effects; third, programme success may be contingent on recognising and addressing cross-dimensional feedback loops. Understanding these cross-dimensional feedback effects may better equip policymakers to approach policy design as a process of feedback management rather than problem solving. This requires certain skills and suggests a capacity for meta-governing through implementation, but it may better support effective governance of complex public sector systems. Methodologically, we have highlighted that approaches to understanding policy success can be enhanced by applying methods that are sensitive to complexity and temporality. Nesting an analysis of policy feedback and related concepts from policy feedback theory within McConnell’s three dimensions of policy success provides one approach to realising this. This paper has illustrated some of the advantages of this. In particular, tracking feedback dynamics within and across dimensions can reveal and provide a framework to explore countervailing forces shaping prospects for policy success in evolving ways. It invites a broader engagement with the idea of success that moves away from a snapshot assessment towards a nuanced, temporally and contextually sensitive assessment of success as an expression of recursive policy dynamics.