J. Cirone1, P. Bendix1, G. An1 1University Of Chicago,Surgery,Chicago, IL, USA
Introduction:
Prior exposure to violence is a known predictor for subsequent interpersonal violence (IPV). Violence recovery programs (VRPs) reduce IPV among high-risk individuals using multifactorial, case management approaches (1), however, little is known of the contribution of the individual VRP components. System dynamics models (SDMs) are a type of dynamic computational modeling that has shown utility in understanding other complex healthcare processes (2). SDMs represent systems as a series of “stocks” (populations) that are linked by interconnected “flows” (transitions) that can be configured as complex feedback loops. Running a SDM produces changes in the various population levels due to programmed transition rates linking one population type to another. Here, we model the general epidemiologic dynamics of IPV and how a VRP may influence IPV risk and recovery.
Methods:
A SDM was created based on an abstract process model of IPV. The model initially simulates flow between low- and high-risk populations, then through IPV and hospitalization events, a potential for death, and a return to the at-risk population. Risk factors such as prior exposure to violence, gang membership, and education were included in IPV risk and event calculations. We included points at which the interventions of a VRP could influence the transition from high-risk to low-risk populations. Model outputs include: trajectories of population distributions, number of IPV events, hospitalizations, and deaths.
Results:
The VRP SDM was successfully implemented using the System Dynamics Modeler in NetLogo and incorporated the features noted above. Simulation experiments involved parameter sweeps of initial population levels, IPV event likelihood and population transition rates. Initial validation of the VRP SDM was achieved by observing output behaviors consistent with known patterns of IPV. Simulation runs converged to stable steady states with the greatest effect on IPV produced by varying the transition propensity between high- and low-risk populations. The VRP also functioned in a recognizable fashion, producing the greatest effect in reducing IPV events by increasing the shift from high- to low-risk populations.
Conclusion:
This initial implementation of the VRP SDM produced recognizable baseline behavior while incorporating the possible effects of a VRP. The VRP SDM will allow us to compare hypotheses of the epidemiology of IPV and evaluate the components of a VRP intervention. Future work will emphasize adding complexity to the VRP SDM and identifying real-world metrics to aid in testing, validation and prediction of the model.
References:
1. Cooper C, Eslinger DM, Stolley PD. Hospital-based violence intervention programs work. Journal of Trauma. 2006;61(3):534-537.
2. Homer JB, Hirsch GB. System Dynamics Modeling for Public Health: Background and Opportunities. American Journal of Public Health. 2006;96(3):452-458.