Professor Kara Hanson reflects on her experiences at RESYST's P4P workshop in Tanzania.

Last week RESYST, together with the Ifakara Health Institute, CMI and the Institute of Tropical Medicine in Antwerp, co-hosted a workshop to bring together programme managers, policymakers and researchers to explore the health system effects of pay for performance (P4P) schemes.

P4P, in which health workers or facilities are paid incentives for providing specific services, is being introduced in many countries, however, evidence of their impact (both positive and negative) is weak.

Over three days, 103 participants discussed and debated a range of country experiences with P4P, for a frank and transparent conversation that (I felt at least), would not have been possible a few years ago when the debate about P4P was more polarized and ideological. 

Here I set out the 10 things I learned (or learned more) about P4P.

1. “The same is different”: Dimitri Renmans from the Institute of Tropical Medicine in Antwerp reminded us that P4P programmes take a wide variety of forms in different settings, underpinned by differences in their theory of change, and influenced by varied political and contextual circumstances. Each evaluation of a P4P scheme needs to take into account this diversity. 
2. P4P alone cannot build the health system: it adds incremental resources, tied to the production of health service outputs, but needs a foundation of health system support on which to build. Where basic inputs (staff, drugs, vaccines) are not present, health workers will be pressured by the new system to perform without the resources to do so, and system change cannot occur. This was the experience of Chad, recounted by Joël Arthur Kiendrébéogo from the Centre Muraz, where the programme faltered after an initial pilot.
3. In most settings donors have driven the process and provided the resources for P4P, but for such programmes to be sustained it will require Ministries of Finance to commit domestic resources. In some settings (eg. Rwanda) the government is contributing some of the funding for P4P, and this share needs to be gradually increased in order for P4P to be institutionalised and sustained.  
4. Ministries of Finance must also be engaged in discussions about how health facilities are paid – and this should happen early in the process.  P4P is a new form of purchasing arrangement, and the financial accounting and audit systems to support it need to be revised to accommodate the change. This needs buy-in and support from the Ministry of Finance.
5. As currently implemented in many countries, P4P may complicate purchasing arrangements by adding a new purchaser (eg a district committee in Rwanda) on top of the existing array of purchasers that health facilities engage with – the Ministry of Health, for its inputs; a health insurance agency for specific services; vertical programmes; etc.  Each will carry its own set of incentives and accountability procedures. In the long run, efficiency gains will result from greater pooling of these different funds, reducing transaction costs.
6. Injecting resources that health facilities can manage directly can positively impact health facility performance even without explicit performance conditionalities, as Evelyn Waweru from KEMRI-Wellcome Trust reported about the Kenyan experience of the Health Sector Services Fund (HSSF). HSSF is a financing mechanism where the Ministry of Health directly transfers funds to health facility bank accounts and facilities have the flexibility to allocate the funds to their determined priorities. However, common to both conditional and unconditional facility-based financing are the challenges of financial management – this is yet another reason to bring Finance ministries to the table.  
7. Quality of care was the elephant in the room: across a range of settings P4P has been demonstrated to improve the quantity of health services, but without clear improvements in service quality. Erik Josephson (Thinkwell) presented a review of indicators used in performance-based payment programmes which showed that 77% of all indicators relate to structural quality (the availability of inputs), and 19% examine processes of care (primarily clinical competence and provider effort). These are important components of quality of care but they do raise the question of whether other, harder to observe clinical processes that are not incentivised will be neglected. There are also questions about the link between structural measures of quality used in P4P evaluation and health outcomes. P4P programmes need to link up with other quality improvement initiatives operating in country to devise ways of improving provider competence (what they know) and ways of getting them to apply what they know in clinical practice.
8. It seems natural to assume that P4P works by influencing health worker motivation, yet most research on P4P skips the motivation step and measures the effects of P4P on health service delivery (Gaute Torvisk, University of Bergen). There were lively discussions about different ways to conceptualise and measure motivation, and we recognised the importance of integrating a broader range of disciplines, particularly behavioural economics and psychology, into evaluation teams in order to better understand the changes in health worker behaviour that are stimulated by P4P.
9. We are still quite confused (or maybe we hold different views?) about the distinction between P4P as a provider payment mechanism, indicating a shift from input-based financing to one in which outputs are reimbursed; and P4P as a way of motivating individual providers, through bonuses paid if targets are reached. In the long run, it seems infeasible to be incentivising individuals to provide all the different services that are available in health facilities.  So to really contribute to a shift in health system models, P4P will have to morph into a provider payment mechanism that operates alongside a number of other mechanisms to balance high- and low-powered incentives
10. With the wealth of experience that is accumulating on P4P in sub-Saharan Africa, it is a great moment for “second generation” programmes to learn the lessons of the early programmes and develop better designs that will avoid some of the lacunae they faced. Researchers also need to develop second generation evaluations that move beyond simple measurement of impact and dig deeper into the causal mechanisms that make P4P work, and also into the political economy issues that will influence the institutionalisation and sustainability of these programmes.

RESYST has funded new analysis of data collected by the Ifakara Health Institute and the London School of Hygiene and Tropical Medicine from pilot of P4P in Pwani Province of Tanzania. This work has looked at the effects of P4P on the availability of resources in health facilities, out-of-pocket expenditure by pregnant women, and accountability mechanisms and relations between stakeholders at different levels of the system.