Research evidence does not transfer as such – transfer requires active network work
Evidence-based practices and policies have become a central guiding principle in health and social care. The idea is appealing: we study what works and implement it everywhere. Science produces the answers, administration implement them, and citizens benefit. Who could oppose such a rational chain?
The problem is not the value of evidence. The problem lies in the worldview on which the model is built.
The linear evidence-based model rests on the assumption that interventions are clearly bounded entities. They can be separated from their environment, tested under controlled conditions, and shown to be effective. Once the effect has been demonstrated, the intervention can be transferred to another context where it is assumed to produce the same result, provided that implementation is done correctly. In this model, the world appears relatively stable: context is a background variable, not an active part of action. Effectiveness is a property of the intervention.
This is a strong claim, as it takes a stance on how reality is understood and what kind of steering is seen as possible.
A relational perspective on the health and social care system
In health and social care, reality is not a collection of separate objects. It is a dense network of people, professional practices, information systems, funding models, spatial arrangements, metrics, legislation, time pressures, and moral expectations that shape one another. An intervention does not enter this world as an external technical solution. It becomes entangled with it, changes it – and is itself changed.
Example: a psychosocial intervention in practice
Take, for example, a psychosocial intervention such as a brief therapeutic treatment model for depression, the effectiveness of which has been demonstrated in randomized trials.
In the research setting, the intervention is precisely delineated: trained therapists, a defined method, a specific number of sessions, and a selected patient group. The result is a statistically significant reduction in symptoms.
When the same model is introduced into a wellbeing services county, however, it does not encounter an empty space. It encounters recruitment difficulties, varying levels of training, busy appointments, information systems that do not support treatment monitoring, funding mechanisms that reward throughput, and clients whose life situations are more complex than in the research setting. The relationship between therapist and client takes shape differently in different places. The “core” of the treatment model does not remain intact but begins to evolve.
If the results are weaker than in the original study, the discussion often turns to implementation: training was insufficient, the method was not followed closely enough, commitment was lacking. More rarely is it asked whether the very idea of transferable, context-independent effectiveness is sustainable.
Effectiveness is an achievement of a network
From a relational perspective that focuses on relationships between things, effectiveness is not an internal property of an intervention but emerges in relationships. It arises from how the treatment model connects to the local division of labour, how leadership prioritises time, how clients recognise themselves as the target group of the service, and how information systems make work visible or invisible. Effectiveness is an achievement of a network, not a property of a single method.
This has political consequences. The linear evidence-based model promises decision-makers manageability: we invest in this model because research shows it works. Responsibility appears clear and measurable. But if effectiveness is relational, governance is not a technical issue but a continuous negotiation. Scaling up is not copying but reconstruction. Every implementation is a change in the entire network.
When policy relies on a linear model, it easily produces the illusion of universal solutions. It overlooks local differences and narrows professionals’ agency to the implementation of standards. At the same time, it shifts risk to the field: if results do not emerge, the fault lies in implementation, not in the model.
What does this mean in practice?
A relational approach does not mean that everything depends on the operating environment (relativism), nor that “nothing is evidence”. It means that evidence is understood as context-dependent and that transferring it requires active network work.
In practice, this means that health and social care operations invest more in local co-development, continuous learning, and the identification of structural barriers. Evaluation does not focus solely on outcomes but also on the kinds of relationships and dependencies that an intervention produces.
Perhaps the most important shift is this: we do not only ask what works, but in what relationships and under what conditions something begins to work. This question is politically less comfortable, because it does not offer quick, universal solutions. But for that very reason, it is a more honest description of the reality of health and social care.
If we want sustainable reforms, we must dare to give up the promise of linear governance. Evidence does not move as such. It takes shape – and we must take shape with it.