Abstract
BACKGROUND: Implementation frameworks and theories acknowledge the role of power as a factor in the adoption (or not) of interventions in health services. Despite this recognition, there is a paucity of evidence on how interventions at the front line of health systems confront or shape existing power relations. This paper reports on a study of actor power in the implementation of an intervention to improve maternal, neonatal and child health care quality and outcomes in a rural district of South Africa. METHODS: A retrospective qualitative case study based on interviews with 34 actors in three 'implementation units' - a district hospital and surrounding primary health care services - of the district, selected as purposefully representing full, moderate and low implementation of the intervention, some three years after it was first introduced. Data are analysed using Veneklasen and Miller's typology of the forms of power - namely 'power over', 'power to', 'power within' and 'power with'. RESULTS: Multiple expressions of actor power were evident during implementation and played a plausible role in shaping variable implementation, while the intervention itself acted to change power relations. As expected, a degree of buy-in of managers (with power over) in implementation units was necessary for the intervention to proceed. Beyond this, the ability to mobilise collective action (power with), combined with support from champions with agency (power within) were key to successful implementation. However, local empowerment may pose a threat to hierarchical power (power over) at higher levels (district and provincial) of the system, potentially affecting sustainability. CONCLUSIONS: A systematic approach to the analysis of power in implementation research may provide insights into the fate of interventions. Intervention designs need to consider how they shape power relations, especially where interventions seek to widen participation and responsiveness in local health systems.