Implementing "from here to there": A case study of conceptual and practical challenges in implementation science.

There is a significant challenge in global health and development research that pivots on the difficulties of delivering (cost-)effective treatments or interventions that are scalable andtransferable across settings. That is, how does one deliver "true effects", proven treatments, into new settings? This is often addressed in pragmatic trials or implementation research in which one makes adjustments to the delivery of the treatment to ensure that it works here and there. In this critical analytical review, we argue that the approach mis-characterises the cause-effect relationship and fails to recognise the local, highly contextual nature of what it means to say an intervention "works". We use an ongoing randomised controlled trial (RCT)-an informal settlement redevelopment intervention in Indonesia and Fiji to reduce human exposure to pathogenic faecal contamination-as a vehicle for exploring the ideas and implications of identifying interventions that work in global health and development. We describe the highly contextualised features of the research and the challenges these would pose in attempts to generalise the results. In other words, we detail that which is frequently elided from most RCTs. As our critical lens, we us the work of American philosopher, Nancy Cartwright, who argued that research produces dappled regions of causal insights-lacunae against a backdrop of causal ignorance. Rather than learn about a relationship between a treatment and an outcome, we learn that in the right sort of context, a treatment reliably produces a particular outcome. Moving a treatment from here to there becomes, therefore, something of an engineering exercise to ensure the right factors (or "shields") are in place so the cause-effect is manifest. As a consequence, one cannot assume that comparative effectiveness or cost-effectiveness would be maintained.