BackgroundWe have conducted an intervention study aiming to improve hospital care for children and newborns in Kenya. In judging whether an intervention achieves its aims, an understanding of how it is delivered is essential. Here, we describe how the implementation team delivered the intervention over 18 months and provide some insight into how health workers, the primary targets of the intervention, received it.MethodsWe used two approaches. First, a description of the intervention is based on an analysis of records of training, supervisory and feedback visits to hospitals, and brief logs of key topics discussed during telephone calls with local hospital facilitators. Record keeping was established at the start of the study for this purpose with analyses conducted at the end of the intervention period. Second, we planned a qualitative study nested within the intervention project and used in-depth interviews and small group discussions to explore health worker and facilitators' perceptions of implementation. After thematic analysis of all interview data, findings were presented, discussed, and revised with the help of hospital facilitators.ResultsFour hospitals received the full intervention including guidelines, training and two to three monthly support supervision and six monthly performance feedback visits. Supervisor visits, as well as providing an opportunity for interaction with administrators, health workers, and facilitators, were often used for impromptu, limited refresher training or orientation of new staff. The personal links that evolved with senior staff seemed to encourage local commitment to the aims of the intervention. Feedback seemed best provided as open meetings and discussions with administrators and staff. Supervision, although sometimes perceived as fault finding, helped local facilitators become the focal point of much activity including key roles in liaison, local monitoring and feedback, problem solving, and orientation of new staff to guidelines. In four control hospitals receiving a minimal intervention, local supervision and leadership to implement new guidelines, despite their official introduction, were largely absent.ConclusionThe actual content of an intervention and how it is implemented and received may be critical determinants of whether it achieves its aims. We have carefully described our intervention approach to facilitate appraisal of the quantitative results of the intervention's effect on quality of care. Our findings suggest ongoing training, external supportive supervision, open feedback, and local facilitation may be valuable additions to more typical in-service training approaches, and may be feasible.