ABSTRACT
Health systems in many African countries are failing to provide populations with access to good quality health care. Morbidity and mortality from curable diseases such as malaria remain high. The PRIME trial in Tororo, rural Uganda, designed and tested an intervention to improve care at health centres, with the aim of reducing ill-health due to malaria in surrounding communities. This paper presents the impact and context of this trial from the perspective of community members in the study area. Fieldwork was carried out for a year from the start of the intervention in June 2011, and involved informal observation and discussions as well as 13 focus group discussions with community members, 10 in-depth interviews with local stakeholders, and 162 context descriptions recorded through quarterly interviews with community members, health workers and district officials. Community members observed a small improvement in quality of care at most, but not all, intervention health centres. However, this was diluted by other shortfalls in health services beyond the scope of the intervention. Patients continued to seek care at health centres they considered inadequate as well as positioning themselves and their children to access care through other sources such as research and nongovernmental organization (NGO) projects. These findings point to challenges of designing and delivering interventions within a paradigm that requires factorial (reduced to predictable factors) problem definition with easily actionable and evaluable solutions by small-scale projects. Such requirements mean that interventions often work on the periphery of a health system rather than tackling the murky political and economic realities that shape access to care but are harder to change or evaluate with randomized controlled trials. Highly projectified settings further reduce the ability to genuinely 'control' for different health care access scenarios. We argue for a raised consciousness of how evaluation paradigms impact on intervention choices.