Abstract
BACKGROUND: There is a growing interest in using pay-for-performance mechanisms in low and middle-income countries in order to improve the performance of health care providers. However, at present there is a dearth of independent evaluations of such approaches which can guide understanding of their potential and risks in differing contexts. This article presents the results of an evaluation of a project managed by an international non-governmental organisation in one district of Pakistan. It aims to contribute to learning about the design and implementation of pay-for-performance systems and their impact on health worker motivation. METHODS: Quantitative analysis was conducted of health management information system (HMIS) data, financial records, and project documents covering the period 2007-2010. Key informant interviews were carried out with stakeholders at all levels. At facility level, in-depth interviews were held, as were focus group discussions with staff and community members. RESULTS: The wider project in Battagram had contributed to rebuilding district health services at a cost of less than US$4.5 per capita and achieved growth in outputs. Staff, managers and clients were appreciative of the gains in availability and quality of services. However, the role that the performance-based incentive (PBI) component played was less clear--PBI formed a relatively small component of pay, and did not increase in line with outputs. There was little evidence from interviews and data that the conditional element of the PBIs influenced behaviour. They were appreciated as a top-up to pay, but remained low in relative terms, and only slightly and indirectly related to individual performance. Moreover, they were implemented independently of the wider health system and presented a clear challenge for longer term integration and sustainability. CONCLUSIONS: Challenges for performance-based pay approaches include the balance of rewarding individual versus team efforts; reflecting process and outcome indicators; judging the right level of incentives; allowing for very different starting points and situations; designing a system which is simple enough for participants to comprehend; and the tension between independent monitoring and integration in a national system. Further documentation of process and cost-effectiveness, and careful examination of the wider impacts of paying for performance, are still needed.