In public health, simulation models are important tools for policy evaluations, and the nature of the problems these models address is complex. System dynamics (SD) modelling is acknowledged as a methodology for complex systems, but there is limited insight in the development and application of SD models for noncommunicable diseases (NCDs) and lifestyle-related risk factors. Therefore, this review investigates how these SD models are developed and utilised to project health outcomes and evaluate policy interventions. Relevant studies were identified through a systematic search strategy. Studies were selected based on eligibility criteria, and data were extracted in two phases: (1) extraction of study characteristics, and (2) detailed examination of applied methods for a subset of included studies focussing on mental health and substance use disorders. For the second phase, data extraction was stratified according to steps of the modelling cycle, and the reporting quality of each study was assessed. In phase 1, 63 studies were included, mainly on mental health and substance use disorders (27%), obesity (17%), and diabetes (13%), with the number of studies increasing over time. In phase 2, 17 studies on mental health and substance use disorders were included. These studies commonly used dedicated software for SD modelling for the mathematical formulation and computerisation of their models. However, in 76% of these studies, the underlying mathematical equations were not shared, and in 82%, the models were not openly accessible. A (slight) majority of studies validated model outcomes against empirical data (65%), and performed sensitivity analysis (53%). Few studies reported model verification (18%) and face validation of model outcomes (18%). SD models are increasingly developed for NCDs and lifestyle-related risk factors, and particularly applied to mental health and substance use disorders. This review offers insights into applied techniques for developing SD models for mental health and substance use disorders, and shows that steps of the modelling cycle are reported on with considerable variation. Although studies frequently report on certain aspects of testing and validation, many report incompletely on other activities of the modelling cycle, and the models sometimes lack transparency. Improving this would enhance the credibility of SD modelling in public health.
Valkenburg et al. (Fri,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: