Performance-based Financing versus “Unconditional” Direct Facility Financing - False Dichotomy?

ABSTRACT A debate about how best to finance essential health care in low- and middle-income settings has been running for decades, with public health systems often failing to provide reliable and adequate funding for primary health care in particular. Since 2000, many have advocated and experimented with performance-based financing as one approach to addressing this problem. More recently, in light of concerns over high transaction costs, mixed results and challenges of sustainability, a less conditional approach, sometimes called direct facility financing, has come into favor. In this commentary, we examine the evidence for the effectiveness of both modalities and argue that they share many features and requirements for effectiveness. In the right context, both can contribute to health system strengthening, and they should be seen as potentially complementary, rather than as rivals.


Introduction
Performance-based financing (PBF) has been proliferating in low-and middle-income countries (LMICs) over the past decade in the health sector, with an increasing number of programs being tested and rolled out, varying in size from small pilots to national scale programs, and supported by a mix of international and national funding. While within the broader results-based financing field, there are diverse models in operation, including conditional transfers in kind, performance-based incentives to health workers, performance-based contracts, results-based aid and target payments, here we focus on the most common model of PBF, which accounted for 37 out of 59 studies in the recent Cochrane review, 1 with the Health Results Innovation Trust Fund alone supporting 35 programs in 29 countries, of which the majority use PBF models. 2 PBF targets performance at health facility level and specific rules on how funds may be used are usually elaborated, including cases where funds may "trickle down" to staff internally, resulting in staff receiving a proportion of funds as incentives. Payments are often based on fixed amounts per targeted service delivered, modified by quality assessments and sometimes other weightings related to equity (e.g., remoteness). 3 Regular verification of the services provided (quantity) and of their quality is conducted in facilities, and this forms the basis for the calculation of facilities' retrospective payments. The evidence base on paying for performance in health care has also expanded exponentially over the last decade, with admissible impact assessments in LMICs growing from 9 to 59 between the Cochrane reviews of 2012 and 2021. 1,4 At the same time, there has been an expansion of research into the underlying theory of change of PBF, 5 its mechanisms of change, 6 its deployment in different contexts, 7 and its systemic 3 and equity effects. 8 While less clearly defined in the literature and less studied, "unconditional" direct facility financing (DFF) has (re)emerged in the debate on facility funding modalities. Under DFF, payments are made directly from a national fund to health care facilities (usually based on prospective approaches such as capitation payments or facility budget plans). 9 As with PBF, these funds are commonly used to finance smaller non-salary recurrent operating expenditures, such as facility operating costs and supplies, with the bulk of facility input costs (salaries, capital expenditure and medicines) being funded separately or provided in kind.
With mixed evidence in relation to PBF's utilization and health outcome effects, 1 concerns about high transaction costs 10 and challenges of integration and sustainability 11 of PBF, a debate has been growing amongst development partners and funders over whether a DFF approach might be more simple, cost-effective and sustainable, while also achieving the goal of increasing resources at provider (especially primary care) level. In this commentary, we consider how PBF and DFF differ, what they have in common, and highlight their shared prerequisites for effectiveness. Rather than being seen as alternatives, they may more appropriately be seen as complementary and must be tailored to each context, ensuring that health financing arrangements in the system overall are moved in the right direction in relation to underlying principles for strengthening health financing. 12

Addressing Shared Challenges
In low-and lower-middle-income settings, PBF and DFF are driven to a large extent by similar fundamental health system challenges. Since the 1980s, governments have struggled to provide adequate and reliable funding to primary facilities in particular, leading to decades of reliance on user charges for small but essential operating costs, with detrimental effects on supply and demand sides, including unmet health needs, inequity of access, poor coverage with essential services and low quality of care. 13 This failure reflects a combination of resource shortages, political economy (which favored higher level facilities), perverse incentives for providers, system failures (such as weak public financial management systems) but also low levels of trust, which challenged any increase in facilities' autonomy. 14

Differences and Similarities in Approach
There are differences between PBF and DFF: in broad terms, PBF funding is tied to specific outputs and quality measures, and so typically has tighter systems of verification. This reduces fiduciary risks, but increases costs and the risks of operating a parallel information and public finance management system. By contrast, DFF operates on a higher trust basis, typically using routine reporting and auditing systems, and therefore may be more appropriate, in general, in more mature or stable health systems. 15 Underlying this is a difference in framing, with PBF focusing more on the importance of effort, incentives, and alignment of principals and agents. In turn, DFF focuses more on the importance of delivering resources, without which providers cannot be held accountable for service delivery. To some extent, these link to ideological positions, as well as reflecting differing professional lenses, which explains some of the heat in the debate, on PBF in particular. 16 Labels are often misleading, however, and the implication that one payment system is dependent on "performance" and the other is not does not reflect the reality. All payments are conditional, although the assessment of performance can be harder (a fixed payment per quality-adjusted output, for example, for many PBF schemes) or softer (for example, payments linked to timely reporting, or meeting agreed facility targets).
Moreover, while the labels direct us to think of finance as a key mechanism of change, all evidence to date (and knowledge of health systems as complex adaptive systems) points to the recognition that PBF and DFF are part of a much broader package of changes, in which other components may indeed be more significant in explaining results (positive, neutral and negative, also unintended as much as intended). 17 For example, both approaches may be accompanied by increased feedback to health staff on their collective performance, greater signaling of priorities, support for planning, giving greater weight to reporting, engaging communities more, and increasing the autonomy to direct resources as needed at the facility level. Each of these have independent and combined potential to improve health service delivery and outcomes.

Shared Pre-requisites
While PBF is sometimes portrayed as complex and DFF as a simple approach, both require considerable groundwork to be effective. For example, both typically require reinforcement of management and accounting skills and capacity at facility level (including in budget preparation and management), as well as improved supervision from the district, investment in robust information systems, and the establishment of facility bank accounts to receive payments. These have been lacking in many LMIC contexts, where financial resources have been managed at the district, or equivalent, level on behalf of facilities. More broadly, neither PBF nor DFF will work in an environment where the national level is unwilling to decentralize decision space and resources, as has been the case in some settings. 18 In addition, there are a number of important program design and implementation components which both approaches need to elaborate, such as estimating resource requirements ("prices" for indicators in the case of PBF, and for DFF the funding amounts needed by facilities, taking into account the degree of subsidies from other sources); determining reporting, verification and performance review approaches; agreeing, monitoring and enforcing policies on charges to users; and determining and enforcing any rules on staff benefits from the funds, and on how funds can be used more generally. Both may be accompanied by donor pooling of funding, thus functioning as a donor harmonization tool to some extent. 15,19 They have also faced shared implementation challenges, such as delays in funds reaching facilities and need for ongoing supervisory support to facilities. 20,21 e2006121-2 S. WITTER ET AL.

Results to Date
More research has been done on the impact of PBF in LMICs to date than on DFF. However, the evidence for PBF's effectiveness remains mixed and the quality of evidence is still assessed as low in general. 1 Compared to the "status quo" (with no additional financing) and for targeted PBF indicators, a recent systematic review 1 found that some utilization indicators (such as provision of HIV testing and prevention of mother-to-child transmission of HIV) had moved positively, while others (such as provision of antiretrovirals and bed net use) had moved negatively, and some were uncertain (e.g., tuberculosis (TB) adherence). There was limited evidence for impact on health outcomes, but some positive evidence for TB treatment success. In general, quality of care (especially structural), resource availability and patient satisfaction were positively influenced, with some exceptions. However, other areas which are important to the theory of change of PBF did not apparently respond so well; these included staff-related indicators (satisfaction, motivation and absenteeism), user fee payments, facility governance and improved equity of utilization. In the minority of studies where unintended consequences were assessed, no major distortions were found. 1 For DFF, more evidence is needed but preliminary findings are somewhat similar to those on PBF, suggesting responses in utilization and quality of care but less impact on user fees to date. 15 In studies comparing PBF with interventions using less conditional financing (which were constructed as comparators for PBF, and did not represent realistic forms of DFF), neither model came out as consistently superior and most evaluations noted that the two interventions had similar impacts overall. When adjusted for additional resources (in five studies), PBF performed somewhat better than controls for some quality and autonomy measures, but less well for utilization. There was no difference in health outcomes, despite slightly higher expenditure in PBF sites. 1

Potential as Health System Strengthening Intervention
Both PBF and DFF should be seen as potential health system strengthening interventions, not just health financing interventions, as they potentially impact all system areas and should in principle be coherent with arrangements in them. For example, they need to take into account: • Existing health worker distribution and remuneration (its adequacy and existing staff incentives and performance management), • Drug supply systems (the reliability, quality and cost of supplies which facilities already access), • Governance structures (including the degree to which the community is engaged in facility management), • Public financial management systems (for budgeting, channeling and accounting for funds, including the degree of financial and managerial autonomy), as well as adequacy and reliability of overall fund flows, • Health information systems (for reporting, verifying performance and identifying bottlenecks), • Service packages (to prioritize essential services and support quality of care), • Infrastructure quality and distribution (as financing a badly distributed network can reinforce inequities and inefficiencies, and facilities need access to IT services, banking, and communications), and • Measures to address community access barriers (as both approaches work mainly on strengthening the supply side).

Conclusions
Whether it is best to address the challenges facing primary facility financing through PBF, DFF, or reforms to basic public financial management systems, or all of them combined, will depend on the context. In settings where basic financing is in place (typically middle-or high-income countries), PBF may play a different role, focusing on marginal incentives for quality. However, in LMIC settings, and especially in fragile settings where many of the programs have been implemented, 7 PBF has provided core finance for recurrent costs at primary facilities and often augmented inadequate salaries for staff. 22 The interaction with donors in LMIC settings has also been significant, with some donors seeing PBF in particular as an appealing approach to make a clearer link between their investments and results. At a higher system level, PBF programs tend to promote separate purchasing, fund-holding and verification agencies, which likely reduce some risks but also increase costs and increase dependence on external organizations. Coherence with wider reform plans is important: if the country is moving toward a purchaser/provider split and individual-based coverage, for example through health insurance, then a PBF model may develop some of the needed capacities, 23 while DFF is more suited to an integrated purchasing model. In many countries, mixed provider payment systems have been found best to manage the advantages and disadvantages of any single approach, and in the longer term, LMICs will likely move to systems that mix predictable base payments, which fit to known population distribution and needs, with a smaller performance-related element. In that sense, DFF can be foundational for PBF, with DFF funding general service provision and PBF focusing on incentivizing utilization or quality for specific services that are identified as suboptimal-or indeed vice versa, where PBF has helped to establish the prerequisites for DFF (not just banking but also planning and financial management capacities at health facility level and supportive national public finance management systems). Verification should be embedded in every payment mechanism, but has to be cost-effective and based on national auditing and observed risks (e.g., analysis of data for outliers, which prompt investigation and controls, which is more cost-effective than the universal verification which many PBF schemes practice). 24 More generally, as a system strengthening intervention, both have promise if designed with good fit to the context and its health system blockages. Both (independently or as a package) can provide the small but essential flexible resources which are needed at facility level to support integrated care packages. Both have the potential to provide a mechanism for donor harmonization. PBF has benefited from a period of intense experimentation and documentation of its model. DFF benefits from a more integrated approach, with lower transaction costs and potentially more sustainability. Both require complementary interventions at facility and community levels. In relation to financial barriers for users, both programs could potentially reduce these, but this component needs more explicit attention and enforcement as results have been disappointing to date.