The study assessed the interactions and the impact of specialist mobile community care teams (assertive outreach teams or AOTs) implemented in the mental health (MH) system of Bizkaia (Spain) using a methodology derived from an ecosystem perspective.

First, the experts assessed the system’s services and codified them according to an international classification system. Second, following an iterative methodology for expert-knowledge elicitation, a clients’ flow diagram showing the inter-dependencies of the system’s components was developed. It included variables and their relationships represented in a causal model. Third, the system elements where the AOTs had a major impact (stress nodes) were identified. Fourth, three scenarios (variable combinations representing the ‘stress points’ of the system) were modelled to assess its relative technical efficiency (technical performance indicator).

The classification system identified the lack of fidelity of the AOTs to the original assertive community treatment model, categorizing them as non-acute low-intensity mobile care. The causal model identified the following elements of the system as ‘stress nodes’ in relation to AOT: users’ families; social services (outside of the healthcare system); acute hospitals; non-acute residential facilities and, to a lesser extent, acute hospital day care services. When the stress nodes inside the healthcare system were modelled separately, acute and non-acute hospital care services resulted in a large deterioration in the system performance, while acute day hospital care had only a small impact.

The development of the expert-knowledge-based causal model from an ecosystem perspective was helpful in combining information from different levels, from nano to macro, to identify the components in the system likely to be most affected by a potential policy intervention, such as the closure of AOTs. It was also able to illustrate the interaction between the MH system components over time and the impact of the potential changes on the technical performance of the system. Such approaches have potential future application in assisting with service planning and decision-making in other health systems and socio-economic contexts.