Community-Centered Interventions Using System Dynamics

Community-centered interventions that are multilevel, multi-setting, and multifaceted show promise in the prevention/early intervention of complex public health problems among youth such as mental health and obesity. System Dynamics (SD) may provide the insight necessary to analyze preventive interventions that adequately match the complexity of problems to the capacity of school/community systems.

We use two empirically-supported, community-centered, school-based preventive intervention approaches (i.e., Positive Behavior Intervention Supports/”PBIS” and Shape Up Somerville. Eat Smart. Play Hard/”SUS”) as case studies. Principles of complex systems analysis were used to conceptualize and guide the process of SD model-building resulting in conceptual causal maps (qualitative) embedded within an ecological systems model. Using a modular subsystem approach, facets of the causal system maps were then separated to permit more visibility in the interactions and feedback between subsystems.

Using this subsystem approach allowed us to begin to develop a quantitative SD model by constructing the form of functional relations and assigning parameter values for one component of the obesity preventive intervention (SUS). The modeling exercise for SUS was confined to the mesosystem of the school food environment subsystem, and focused on those children known to attend an after-school program also involved in the intervention. This was done purposefully to allow for interactions between the different actors and child to be realized.

Translating conceptual/causal maps into quantitative SD models is challenging and not typically employed in prevention science—although these methods are emerging. Embedding the SD model within an ecological framework made it possible to trace multiple influences that reinforce mental and physical behavior. Resource flows indicated capacity-building through resources and support.

SD modeling provides several features that are particularly useful for multilevel, multifaceted community-centered program design, implementation, and transfer to other settings.  For instance, once the parameters of the SD model are determined for a particular setting, it would provide insights on capacities at various layers and potential barriers.  Feedback and resulting interactions between subsystems were driven primarily through measurement of outcomes.  In PBIS, it was the degree to which the approach prevented school dropouts, poor classroom behavior and poor academic outcomes, which prompted action in the micro- and mesosystems to increase a supportive school climate. In the SUS example, prevention of unhealthy eating at school resulted in the interaction amongst numerous actors and a process of continuous adaptation to increase the effectiveness of the intervention.

While limitations to SD and this study do exist (and will be discussed), SD modeling shows promise in prevention science for two main reasons: 1) to translate the systems thinking that underlies the design of multilevel, multifaceted community-centered interventions into a quantitative form, and 2) to provide a simulation platform to help the actors in the outer layers visualize the possible influences of their actions.