Abstract
BACKGROUND: Implementing complex and multi-level public health programmes is challenging in school settings. Discrepancies between expected and actual programme outcomes are often reported. Such discrepancies are due to complex interactions between contextual factors. Contextual factors relate to the setting, the community, in which implementation occurs, the stakeholders involved, and the characteristics of the programme itself. This work uses realist evaluation to understand how contextual factors influence the implementation process, to result in variable programme outcomes. This study focuses on identifying contextual factors, pinpointing combinations of contextual factors, and understanding interactions and effects of such factors and combinations on programme outcomes on different levels of the implementation process. METHODS: Schools which had participated in a school-based health promotion programme between 2012 and 2015 were included. Two sets of qualitative data were collected: semi-structured interviews with school staff and programme coordinators; and written documents about the actions implemented in a selection of four schools. Quantitative data included 1553 questionnaires targeting pupils aged 8 to 11 in 14 schools to describe the different school contexts. RESULTS: The comparison between what was expected from the programme (programme theory) and the outcomes identified in the field data, showed that some of the mechanisms expected to support the implementation of the programme, did not operate as anticipated (e.g. inclusion of training, initiation by decision-maker). Key factors which influenced the implementation process included, amongst other factors, the mode of introduction of the programme, home/school relationship, leadership of the management team, and the level of delegated power. Five types of interactions between contextual factors were put forward: enabling, hindering, neutral, counterbalancing and moderating effects. Recurrent combinations of factors were identified. Implementation was more challenging in vulnerable schools where school climate was poor. CONCLUSION: A single programme cannot be suited or introduced in the same manner in every context. However, key recurrent combinations of contextual factors could contribute to the design of implementation patterns, which could provide guidelines and recommendation for grass-root programme implementation.