Moving beyond formulae: a review of international population-based resource allocation policy and implications for Ireland in an era of healthcare reform

Population-based resource allocation and population health approaches to health systems

An important distinction between population health approaches and other approaches to health planning is that they encompass the entire range of determinants of health and wellbeing (e.g. education, housing, town planning) as well as the planning and management of health services⁵. The core aim of PBRA frameworks is to facilitate strategic health service planning by shifting towards models that incorporate a more holistic assessment of needs, rather than focussing on specific diseases, settings or service utilisation patterns. PBRA is used to adjust funding across the population according to variation in need and the cost of providing required services and supports (i.e., cost weightings). PBRA policies are also considered to be crucial to the process of decentralisation since they help to ensure that allocation of healthcare resources is responsive to variation in regional/local population profiles.

PBRA frameworks comprise a core model that estimates the relative proportions of expenditure required to provide services based on a set of pre-defined circumstances. These circumstances should, in theory, reflect healthcare need and explain differences in service utilisation patterns⁴. Measuring population health need is a complex and context-specific process that has been covered extensively in the literature^2,6,7; however, it typically incorporates population size, demographic characteristics and additional factors such as unmet need or variation in cost of supply. There are further issues to consider when implementing a PBRA model, including the level of ‘top slicing’ required, the need for transitional funding arrangements and how resources should be distributed at both the regional- and programme-level. It is worth noting that recent evidence from England has suggested that robust methods at the national- (or regional-) level may not necessarily translate to the local-level to help deliver services (e.g. payments to providers were not aligned with intended integration of services and needs of patients with multiple conditions)⁸.

While there are a number of population health planning methods that can be used to support PBRA implementation - including health needs assessments, population profiling, population segmentation or stratification and population health management - there is no standardised approach to the development and implementation of PBRA funding models. This means there is no consensus about the indicators that should be included in the modelling or the analytical approach that should be employed¹. Nevertheless, PBRA has assumed importance in many health systems because it is viewed as a valuable policy lever for promoting equity in health outcomes and access to services; helping to create incentives within the system to promote service reform; and providing explicit criteria for funding, thus promoting informed dialogue and acceptance of outcomes among stakeholders⁴.

Population-based resource allocation and the reform agenda in Ireland

Established in 2005, the Health Service Executive (HSE) is responsible for the allocation of public resources as well as directly providing many of the health and social care services which it funds⁹. However, current funding arrangements were neither designed to re-orient the Irish health system to primary and community care services, nor to support the delivery of integrated care. Indeed, the lack of PBRA arrangements have been repeatedly critiqued as being a weak point of the healthcare system in Ireland^7,10–12 and although recent legislative developments provided powers to the Minister for Health to direct the conduct of a health needs assessment¹³, processes around linking these data with health service planning remain undeveloped.

Resources are allocated based on existing level of service (ELS) - that is, historic patterns of population demand amongst those who have presented for and received care - rather than being guided by current and projected needs in line with overall policy objectives (though it is worth noting a small but increasing proportion of resources for acute hospital services are allocated based on a case-mix programme)¹². This approach is problematic since utilisation patterns may not accurately reflect population need, particularly where there are cost barriers and long waits to access care, thereby reinforcing structural inequities. Research has also pointed to significant limitations in the quality and availability of data to measure health need in the Irish context¹⁰. As a consequence, Ireland’s health system lags behind those in other high-income countries where PBRA has become a key mechanism for delivering equitable, effective and efficient care⁷.

Ireland has recently embarked on a path of systemic and large-scale healthcare reform. Initiated in 2017, a 10-year programme or roadmap for change - ‘Sláintecare’ – was published with the aim of transitioning to a universal health system that is based solely on need rather than ability to pay². Put simply, it seeks to transform healthcare policy and service-level infrastructure to provide all citizens with the ‘right care’ in the ‘right place’ at the ‘right time’^14,15. The 2017 report recommended the introduction of regions to facilitate population based resource allocation and clinical and managerial governance and accountability¹⁵. While some progress has been made¹⁶ the roll-out of Sláintecare has been characterised as slow¹⁷; nevertheless, the most recent Sláintecare Implementation Strategy and Action Plan 2021-2023 restated a commitment to universal access to integrated care¹⁸. Critically, it also specified two primary reform programmes: 1) improving safe, timely access to care and promoting health and wellbeing; and 2) addressing health inequalities on the journey towards universal healthcare.

In the case of the latter, there are two projects most relevant to PBRA: 1) the development of a Citizen Care Masterplan (CCMP); and 2) the establishment of Regional Health Areas (RHAs). Significantly, three of the work streams associated with developing the CCMP are directly related to population health planning, including: population profiling and segmentation; population needs assessment and service redesign; and the development of a PBRA model. The stated rationale for developing RHAs also includes explicit reference to improving a population-based approach to service planning and the integration of community and acute services. The action plan sets out specific quarterly targets, with a view to test, refine and implement PBRA in 2022 and 2023.

What is clear is that PBRA is intended as a key component of Irish health system planning and policy implementation, signalling that policy-makers view PBRA as instrumental to the goals of Sláintecare. As health system planners begin to consider the options for developing a PBRA framework, there is much to gain from examining how PBRA has been designed and implemented in other jurisdictions with transferrable knowledge for Ireland.

Aims and objectives

This review aims to inform Sláintecare’s strategic policy on healthcare reform by generating insights into what PBRA ‘looks like’ in other countries as well as how it has been experienced across countries via a documentary analysis and rapid review of international policy and practice. We address three research questions (RQs):

    RQ1. How are population-based resource allocation policies specified and realised?

    RQ2. What is known about the implementation or impact of population-based resource allocation policies?

    RQ3. What lessons, if any, may be relevant for Sláintecare implementation in Ireland?

In line with the WHO health systems framework, where financing is one of six key building blocks used as a mechanism to ensure access and coverage to quality safe care, a core goal of this review is to progress thinking beyond the technical aspects of formula(e) construction into a broader health systems frame¹⁹. This review is also cognisant of the complexity of health systems and seeks to situate the findings in the context of a complex adaptive system²⁰. Health systems and policy research continually highlights the importance of context for all of the health system building blocks, including financing and resource allocation^19,21,22. This facilitates deeper understanding of the factors related to model development as well as how, and why, PBRA policies have delivered or not on their key objectives. In so doing, we identify aspects of model design and application that have important implications for Irish healthcare reform and present key learnings that could be useful for Sláintecare implementation during a critical window for transformation.

Study design

This study is one work package within the Foundations’ research project that aims to harness key learnings from Ireland’s health system response to COVID-19 and assess whether these insights can be used to inform Sláintecare implementation (for a detailed overview, see Burke et al.²³). This particular component incorporated two qualitative methods - documentary policy analysis and a narrative rapid review - into an integrated methodology to generate robust findings to address the research questions. The combination of these methods is well suited to this form of inquiry, whereby the data collected and processes described are predominantly text-based and in which depth of detail is important to the audience. To this end, a concurrent multi-method approach was employed to create a fuller ‘picture’ of existing knowledge and understanding of international PBRA policy and practice²⁴. This facilitated a comprehensive and nuanced analysis from which important implications for Sláintecare design and implementation emerged (addressing RQ3).

Sampling

Six jurisdictions are included in this review: Australia (New South Wales), Canada (Alberta), England, New Zealand, Scotland and Sweden (Stockholm). The cases represent a convenience sample and were selected based on the following criteria:

An overview of the health system and population characteristics of the six jurisdictions studied is provided in Table 1 and is presented alongside Ireland for comparative purposes. Healthcare funding in all six countries is predominately funded through general taxation. Private health insurance markets operate in all six, with policies offering supplementary or complementary coverage.

Data collection and analysis

1. Documentary analysis: Policy, strategy and planning documents relevant to the review were collected from relevant databases, websites and input from project partners or local experts. These documents were descriptively analysed using matrices in Microsoft Excel and qualitatively reviewed for content on how PBRA is specified and realised in the jurisdictions selected as case studies (addressing RQ 1). While the level of detail available varied substantially across countries, the following details were extracted from the selected policy documents where possible: country, year, policy context, reported policy aim or purpose and the description of the population-based approach (including reference to segmentation, oversight and evaluation).

Results were extracted into summary tables by one researcher (BMJ) and verified through discussion among team members (BMJ and SB). Once compiled, these tables were developed into country-level summaries. Each summary was reviewed and verified by local experts affiliated with academic institutions or national agencies, to maximise accuracy and completeness of the data.

2. Narrative rapid review: A rapid review of peer-reviewed and grey literature was completed to synthesise evidence on the impact, experience and implementation of PBRA policies across the six countries examined (addressing RQ2). A systematic search strategy was used; however, rather than attempting to identify all evidence, this review focused on synthesising key evidence related to the research questions. All results are reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis Statement (PRISMA).

We did not restrict our search to any one design. The following inclusion criteria were used to identify materials relevant to this review:

A systematic search strategy was developed through discussion amongst the research team and employed in May 2020, using Pubmed (see Box 1). Supplementary searches for evidence were conducted through reference mining of all identified comparative studies and systematic reviews. In addition, Google Scholar was searched for grey literature sources and other studies using keywords from the database search. For Google Scholar, the first 100 items were collated. One team member (BMJ) reviewed each grey literature item for potential relevance to this review. Studies published in the peer-reviewed literature were discarded if they had already been returned by the database search.

References from the database and grey literature searches were managed using EndNote. All references from the database search were uploaded to the online reviewer tool Covidence to screen for relevance using title and abstract. Papers that did not meet the inclusion criteria were excluded at this point. Full texts were reviewed for inclusion by one researcher (BMJ), with all choices confirmed through discussions with another researcher (SB).

We decided a priori to extract data relating to publication date, country, overview of publication and reported findings. Narrative synthesis was conducted whereby all resources were appraised and the relevant findings collated into a coherent account to address the study’s research questions.

Our search of the Pubmed database and grey literature is summarised in Figure 1. A total of 1,562 papers were identified following a database search in Pubmed. An additional 191 resources were identified via searches on Google Scholar, government websites and subsequent reference mining. The total yield was 1,753 citations, of which 41 were duplicates. We reviewed the remaining 1,712 unique titles/abstracts, of which 1,657 did not meet the eligibility criteria. We reviewed the remaining 55 full texts, of which eight met the inclusion criteria. These eight articles were included in the narrative synthesis.

Figure 1. Database search and study selection.

Ethical approval

This study was granted ethical approval by the Research Ethics Committee of the Centre for Health Policy and Management and Centre for Global Health at Trinity College Dublin as part of the Foundations’ Project²³.

Documentary analysis

Table 2 presents a descriptive account of the stated objectives of PBRA policies in the six countries studied. Since there is no standardised framework for PBRA models, the goals of PBRA policy differed across jurisdictions. While all broadly aimed to link resource allocation to population need, there were notable differences in their explicit commitment to policy objectives in official documentation. For example, England cites equity in access as the guiding policy objective, while New Zealand cites equity in health status.

In terms of how the six case studies translated policy into practice, an overview of the PBRA models, services covered and arrangements for funding outside of the models is provided in Table 3. Additionally, details on the specific factors used in population-based funding formulae are outlined in Table 4. Similar to other comparative reviews^3,6, the findings presented in Table 3 demonstrate that all used population size as a starting point for determining resource allocation requirements and then adjusted based on factors that act as indirect measures of need (e.g. age, gender, deprivation) or more direct measures of need (e.g. morbidity); all of which are characteristics considered to influence demand for services. However, Table 4 highlights that some models were further adjusted for differences such as rurality (in the case of Scotland and Alberta, Canada) or included weightings for unmet need. In New Zealand, for instance, the model is adjusted for ethnicity to reduce known health disparities among ethnic-minority groups in the population.

Notably, there was also variation in terms of the stated criteria for choosing the variables to include in PBRA models (Table 5). Since there is no generally accepted working definition of ‘need’, decisions about adjusting PBRA models are largely subjective. It is perhaps unsurprising, then, that different countries used different variables in their modelling to account for population need. As can be seen however, the guiding principles across most countries were that variables should be feasible to include without risk of high administrative costs; be universally recorded, consistent, reliable and verifiable; be able to predict health service expenditure; be free from perverse incentives; and not be a measure of supply or vulnerable to manipulation or fraud.

Returning to Table 3, the findings revealed significant variation in the types of services funded by PBRA models as well as the additional funding arrangements implemented across countries. In some jurisdictions - like Sweden - funding formulae are applied to specific services like primary or acute care, while in others - like New South Wales, Australia - PBRA models fund an explicit health benefits package or bundle of services offered nationally. The variation in services covered probably reflects differences in population entitlements, availability of data, degree of decentralisation and policy priorities. Moreover, across most jurisdictions studied, the arrangements for ‘top slicing’, transitional funding, and interregional or inter-programme transfers were poorly described and often lacked specific details on how models support service planning beyond the allocation of monetary resources.

Finally, a summarised account of the processes for reviewing and revising PBRA policy and model design across the six jurisdictions studied is presented in Table 6. Almost all had established expert advisory groups to review the structure of the formulae and the resource allocation estimates they generate. New Zealand, Scotland and England are examples of countries that publish reports and external audits of funding formulae on a regular basis. A small number described the explicit criteria guiding model development and refinement; however, overall, most policy documents provided limited details about how PBRA models are operationalised and evaluated.

Rapid Review

To develop a deeper understanding about the experience of PBRA, a rapid review was undertaken to summarise recent evidence on the impact and implementation of PBRA approaches in the six jurisdictions covered in this research. Eight relevant studies were selected for inclusion, reporting on PBRA in New Zealand (n=3); Stockholm, Sweden (n=2); England (n=1); Scotland (n=1) and Alberta, Canada (n=1). A comparative review that drew on data from these five regions as well as New South Wales, Australia and the Netherlands was also included in the analysis. Details on the selected studies as well as the key findings identified during the narrative synthesis are presented in Table 7 and are subsequently discussed in relation to three themes: the design, implementation and evaluation of PBRA policy.

Design. Evidence suggests that PBRA approaches have promoted technical efficiency³ in some cases, but that perverse incentives can remain if funding does not align with policy objectives. This emphasises the need to ensure a viable connection between policy goals and model design. In Stockholm, Sweden, for instance, evaluative data indicated that the transition to a fee-for-service approach in 2008 created incentives that were not aligned with the values underpinning the system (that is, equity). Because of this, they pivoted back to a needs-based system which demonstrably improved access for marginalised groups or those with greater health need²⁵. Similarly, Alberta’s model employs a flexible needs-based resource allocation process rather than a hard formula to ensure that their funding model meets the stated aims of the policy, which is to ensure that all regions are able to operate on a more level playing field²⁷.

Implementation. Since, in many countries, PBRA is still in a relatively early development stage, it is perhaps unsurprising that there is generally little evaluative research that moves beyond PBRA design to examine implementation, impact and outcomes. A key problem is the transition from the old allocation system to the new PBRA model, especially where that requires a significant shift in resources. This in turn raises questions as to how quickly to implement the new model and the availability of resources to do so. However, available research indicates that PBRA approaches can help to improve equity in terms of health outcomes²⁸ and access²⁹. In England, specifically, additional funding had a much greater impact in terms of reducing amenable mortality in areas of higher deprivation than in affluent ones²⁸. The findings also suggest that funding models are critical to successful regionalisation policies²⁷ and that variation in efficiency can be influenced by the provision (or lack thereof) of clear guidelines and direction provided by government³⁰. All of this points to a degree of tension regarding decentralisation; that is, between providing structure and allowing for service delivery and planning to be informed by the local context.

Evaluation. As mentioned, robust governance and monitoring mechanisms can play a critical role in helping jurisdictions to assess and measure the extent to which their model design is delivering (or not) on their policy goals²⁵. Importantly, however, even in countries like New Zealand where transparent evaluation processes exist (such as annual reports and external audits), there have been calls for improved accountability and public participation, particularly around model refinement³¹. Notably, civil servants and other stakeholders in Alberta, Canada played a crucial role in developing the design of the PBRA approach implemented, with strong stakeholder engagement being cited as a key factor for ensuring little opposition and a seamless transition to a needs-based funding model²⁷.

Designing population-based resource allocation

The findings suggest that before any technical aspects of the model should be considered, clearly defining the objectives and rationale of the model is key. International evidence suggests that this is critical since the overarching goal will influence the type of model or approach required to achieve it, whether it is equity in healthcare outcomes, adjusting funding to match need or equal opportunity for service regions. Moreover, this goal orientation will also guide and affect all other aspects of implementation, including assessment through ongoing monitoring and evaluation. There is therefore a need for Irish decision-makers to explicitly identify the stated objectives of the model at the outset of policy development beyond simply acknowledging the aim of supporting Sláintecare. Notably, the findings indicate that this process would be bolstered by establishing clarity and standardisation of terminology as well as enabling stakeholder involvement; all of which would help to create trust in a shared vision of the model and foster commitment to partnership. For this reason, consensus and a shared understanding amongst stakeholders about how these aspects of PBRA are articulated and mobilised in Ireland should be a critical first step to maximise outcomes. Whether Sláintecare’s PBRA approach aims to achieve health equity or universal access to care, the rationale should be carefully considered, collaboratively developed and clearly articulated.

The review emphasised that it is also critical is to ensure that policy objectives are aligned with the model design. We acknowledge that there is currently no consensus regarding the ideal framework for PBRA models and that approaches will be context specific insofar as they reflect a mix of technical, practical and political factors. However, evidence from other countries revealed a number of contextual considerations for Sláintecare’s PBRA model development that could help to link the overarching policy goal(s) with model design to maximise outcomes. These include identifying:

Critically, this not only requires access to appropriate universally recorded, regularly updated and reliable data on the factors deemed relevant to modelling health need, but also accurate data on costs to model expected expenditure and any unavoidable differences between regions. This is a complex exercise that, in turn, requires a range of expertise in relevant subject matters (such as health policy, public health, epidemiology and health systems performance) as well as health intelligence and data analysis. Following New Zealand, for example, it could be argued that Ireland’s model should be designed to adjust for ethnicity to reduce known disparities in health outcomes amongst those in the Traveller community and other ethnic minorities. In this instance, it would be important to ascertain what kind of data are available to assist with this; whether these data are accessible and reliable in that they are uniformly and systematically collected; and what the weighting assigned to different types of services should be.

International experience has demonstrated that a multidisciplinary approach to design can help to address these issues and ensure the model is both aligned with the broader objectives of health systems reform and measures performance on related metrics. As such, Irish policy-makers must consider the best way to facilitate partnership working across disciplines and settings to not only define and capture population need, but also identify the services and capacity required to meet that need. Since health services often struggle to develop adequate levels of technical expertise^33,34, initiatives to support improvements in workforce capacity - including education and training, professional development schemes and stimulating demand from leaders - will be crucial to this process.

Implementing population-based resource allocation

A key theme that emerged from the review was the need to support decentralised resource allocation. Population-based funding formulae are regarded as an essential prerequisite of successful decentralisation policies. Having a clear budget, along with appropriate guidance, provides regions with the flexibility required to respond and adapt to local circumstances. With robust outcome measures and data collection in place, regions can also be accountable for their performance, not just in terms of activity and financial control, but also in terms of their contribution to the health status of the populations they serve. However, the literature from New Zealand revealed a tension between national policies and strategies and allowing service planning and/or delivery to be informed by local contexts; while in England, challenges have arisen when resources were allocated to the local levels (or the regions, in the Irish context) but they did not have the autonomy or capacity to utilise them well⁸. Moreover, across all jurisdictions studied, the arrangements for ‘top slicing’, transitional funding, and interregional or inter-programme transfers were poorly described and policy documents typically lacked clarity about the specific mechanisms that can support integrated care in addition to the allocation of funding.

A key learning for Ireland is that PBRA policy is not a panacea for health system reform. In other words, it is not just about providing the ‘right’ proportion of a budget to each region to deliver healthcare services; rather, for PBRA to be effective, other elements necessary as part of a broader programme of health system reform must be in place - such as workforce planning and clarity on an entitlement to universal healthcare - to enable the regions to successfully adapt to their population’s needs⁴. For this reason, it is critical to identify all mechanisms beyond the provision of funding that need to be up and running at the regional level to support the operationalisation and efficacy of PBRA. Put simply, if healthcare is to be delivered via the regions - as envisioned in the Sláintecare Implementation Strategy and Action Plan 2021 - 2023 - key decisions are required about how much healthcare will be delegated to the regions; how much autonomy they will be given, particularly around workforce; and how they will be supported to utilise funding effectively. These decisions need to be made in advance of being able to progress the PBRA model.

Evaluating population-based resource allocation

Finally, the review highlighted the importance of developing robust governance and evaluation mechanisms. PBRA policies are well-established in many countries; however, there are fewer details on the criteria used to evaluate population-based models. This is despite the literature suggesting that it is critical to monitor and assess whether the model is delivering and to adjust as needed to meet the stated policy objectives³⁵. Indeed, the implementation of PBRA is an ongoing, emergent and complex process and, as shown, there is evidence from the Swedish and Canadian experience to suggest that stakeholder engagement or consultation (particularly with the public) is a key mechanism to ensure PBRA approaches are effectively developed over time.

An important implication for Sláintecare is that clarity on governance structures for PBRA in the context of broader health system and policy goals in Ireland will be important; as will be developing from the outset a system for transparent research and reporting on the impact and evaluation of PBRA. Mechanisms for ensuring this can include: the establishment of an expert advisory group to review the structure and outputs of funding formulae; providing detailed information about the criteria guiding model refinement; and/or the publication of reports or external audits on a consistent basis. Since PBRA is a process that should be amended as Ireland’s population profile and needs change over time, important decisions will therefore be required in relation to how the model will be continuously and rigorously monitored, evaluated and refined to allow for future changes. A robust governance and monitoring system will, in turn, enable these learnings to be fed back into the system to maximise health outcomes in the Irish context.

Underlying data

All data underlying the results are available as part of the article and no additional source data are required.