Novel methods for optimising health systems payment for performance interventions to improve maternal and child health in low-resource settings

There are different ways to pay health care providers. They can be paid based on the resources they use (e.g. staff, drugs) and the size of their population. They can also be paid based on what outcomes they achieve, which is termed payment for performance (P4P). P4P schemes are aimed at influencing the behaviour of health workers and their managers to deliver better quality health care. P4P schemes are currently being implemented in many low- and middle-income countries to improve maternal, newborn and child health.

Many studies have focused on assessing these P4P programmes' impact on the performance targets. More recently, studies have considered the effect of P4P on the inputs and infrastructure required to deliver health care services (i.e. the health system). However, these studies have mainly focused on effects on a single element within the health system, e.g. drugs or staff, rather than looking at all these factors in an interconnected, comprehensive manner. Mathematical models enable the analysis of how interconnected systems such as the health system function and respond to change. While models have been built and used to look at health programmes, to date there has been very limited use of these models to study health systems in low- and middle-income countries and their response to programmes such as P4P. One of the advantages of models is that they can also be used to anticipate the likely effects of programme changes before these changes are actually made. This can be helpful to those designing programmes to help them work out which design would work best.

This study will use two types of models to understand the effects of P4P: a model of how a health facility functions in terms of the overall flows of patients and drugs and supplies across the facility (i.e. a system dynamic model) and a model of the way health care workers, their managers and patients interact and behave within the health facility (i.e. an agent-based model). The models will be developed in two different settings: Tanzania and Zambia. Tanzania and Zambia were chosen as they have made mixed progress in terms of maternal, newborn and child health outcomes and they both made the decision to introduce P4P to try to improve maternal and child health. These two models will provide us with an understanding of how health systems work and respond to the P4P programme in these settings, and how P4P can be best designed for maximum impact on the health of mothers, newborns and children. To construct models of the Tanzanian health care system, we will use information from a previous study of the impact of P4P in Tanzania together with interviews with programme implementers. The models will be used in Zambia to see if they can accurately predict the effects of P4P on the Zambian health system or if changes in the model structure are needed. The models will be used to understand the effects (both intended and unintended) of P4P in each country and explore how changes in the design of the Tanzanian and Zambian P4P programmes may affect health and health systems outcomes. Results will be used to improve P4P programme design in each setting. We will also develop a toolkit for how to develop and use system dynamics and agent-based models to analyse health system response to interventions such as P4P for use in other countries, and training activities to support their uptake. This project will support knowledge sharing and learning across partners in the United Kingdom, Tanzania, Zambia and Uganda.

This project will apply novel methods rooted in complexity science to understand and optimise health systems payment for performance (P4P) interventions to maximise impact on maternal, newborn and child health (MNCH) in low-resource settings. P4P is a mechanism to improve the availability, quality and utilisation of essential health services through financial incentives to providers and healthcare managers. It has been applied widely in high-income settings and low and middle income countries (LMIC) have also started its implementation to improve MNCH.

This research focuses on Tanzania and Zambia - a low- and a lower-middle income country, respectively, which have been implementing P4P schemes to improve MNCH outcomes since 2011. This study will build a system dynamic model (SDM) and an agent-based model (ABM) of the Tanzanian health system and its response to P4P. System dynamics represents a top-down approach to model the complex macro behaviour of the health system. ABM, on the other hand, is a bottom-up approach which can model emergent behaviour resulting from interactions between health workers, health care managers and patients. These models will be developed based on in-depth interviews, available data and literature reviews, and calibrated using 85%-90% data from the impact and process evaluations of P4P in each country. We will validate the SDM and ABM in the Tanzanian and compare ABM and SDM in characterising the P4P scheme in Tanzania. We will examine the generalisability of the model to Zambia. We will use these novel models to examine the intended and unintended effects of the P4P scheme that were not previously measured or observed, and to predict the effects of feasible changes to the design of the P4P scheme and examine what changes to the programme design would be needed to reach the desired output/outcome level. We will also build capacity among decision makers and researchers on SDM and ABM to study health systems using models in LMIC.

This research is intended to benefit three main groups: a) national stakeholders in low- and middle-income country settings; b) international donors and agencies investing in health systems; c) the academic and researcher community. In addition, as these groups act on the study's findings, there is clear potential for health workers and their managers to benefit in terms of a better work environment and for patients to benefit in terms of increased access to higher quality and more effective care, which will result in better health. Here, we set out how each group is expected to benefit from the research, and the Pathways to Impact section sets out the strategies for achieving this benefit.

Study findings will be of interest to national governments, payment for performance (P4P) implementers and donors supporting these programmes in Tanzania and Zambia, by helping to better understand how such investments affect health systems and maternal and child health and how to optimise impact from these investments and maximise value for money. The models will be developed with a user-friendly software interface, and can provide the foundation for models considering the effects of other health system strengthening interventions in the future. The methodological toolkit developed can guide other countries on design choices, where to implement and how to monitor effects over time.

This research will interest the international community of donors and agencies financing and supporting the implementation of P4P, as the findings may help to shape their own investments in other countries. Furthermore, they may be able to adapt and use the models in other settings.

The academic beneficiaries and the benefits are outlined in the Academic beneficiaries section.