Impact evaluation and RCTs in health

I am currently working on a proposal for a pilot of performance-based contracts (PBC) in Uganda’s health sector, and have been busy navigating the literature out there. Fortunately for me, there is also a lot of discussion on randomized controlled trials (RCTs) and impact evaluation in the development blogosphere of late.

Today I’ve been reading “Performance Incentives for Global Health,” published by the Center for Global Development, and available for purchase or downloadable chapter-by-chapter here. It has proven very useful so far in helping me think through the various ways in which PBC pilots could be designed. In Chapter 5, A Learning Agenda, the authors write:

Impact evaluation is more than a tool for gauging impacts at the end of a program and providing the inputs into a cost-effective analysis. It can also help a program to evolve. For example, in the initial phase of a pay-for-performance program, three contracts with different risk levels can be piloted. Based on the results from an early evaluation, the most effective contract can be scaled up. Several parameters lend themselves to this kind of experimentation, including the relative effects of supply versus demand interventions, the level of rewards offered for performance, and the balance of trade-offs between access and use.

Rwanda is often noted as a pay-for-performance (P4P) success story, and Chapter 10 is devoted to this case study. The original study by Basinga et al. (2010) is available here (and co-authored by Rwanda’s current Minister of Health, Agnes Binagwaho, also available at @agnesbinagwaho). The authors find that P4P has a significant effect on the number of deliveries in health facilities, quality of prenatal care, and number of preventive care visits for children, but they find no effect on the number of prenatal care visits or immunizations.

In Uganda, a similar pilot, this time of private-not-for-profit (PNFP) facilities, found no effect of bonuses on health facilities’ performance in achieving self-selected health targets. They did find that financial autonomy improved health facilities’ performance, however. The study, “Contracting for Primary Health Care in Uganda”, remains an unpublished World Bank manuscript, as far as I can tell (publication bias at work), but the slides from the 2007 CGD presentation are available here.

I’d like to examine the effect of PBC on health outcomes in the public rather than PNFP sector (hopefully using a few variations of the contract “treatment”), as well as better understand why performance-based pay (in the form of bonuses) did not seem to have an effect on health outcomes in the Uganda pilot. Finally, I am interested in understanding the relative efficacy of supply-side (such as PBC) vs. demand-side (such as conditional cash transfers) efforts in improving various health outcomes. More updates on this to come.

%d bloggers like this: