Applications of systems modelling in obesity research.

Obesity is a complex system problem involving a broad spectrum of policy, social, economic, cultural, environmental, behavioural, and biological factors and the complex interrelated, cross-sector, non-linear, dynamic relationships among them. Systems modelling is an innovative approach with the potential for advancing obesity research. This study examined the applications of systems modelling in obesity research published between 2000 and 2017, examined how the systems models were developed and used in obesity studies and discussed related gaps in current research. We focused on the applications of two main systems modelling approaches: system dynamics modelling and agent-based modelling. The past two decades have seen a growing body of systems modelling in obesity research. The research topics ranged from micro-level to macro-level energy-balance-related behaviours and policies (19 studies), population dynamics (five studies), policy effect simulations (eight studies), environmental (10 studies) and social influences (15 studies) and their effects on obesity rates. Overall, systems analysis in public health research is still in its early stages, with limitations linked to model validity, mixed findings and its actual use in guiding interventions. Challenges in theory and modelling practices need to be addressed to realize the full potential of systems modelling in future obesity research and interventions.

Improving health systems performance in low- and middle-income countries: a system dynamics model of the pay-for-performance initiative in Afghanistan.

System dynamics methods were used to explore effective implementation pathways for improving health systems performance through pay-for-performance (P4P) schemes. A causal loop diagram was developed to delineate primary causal relationships for service delivery within primary health facilities. A quantitative stock-and-flow model was developed next. The stock-and-flow model was then used to simulate the impact of various P4P implementation scenarios on quality and volume of services. Data from the Afghanistan national facility survey in 2012 was used to calibrate the model. The models show that P4P bonuses could increase health workers' motivation leading to higher levels of quality and volume of services. Gaming could reduce or even reverse this desired effect, leading to levels of quality and volume of services that are below baseline levels. Implementation issues, such as delays in the disbursement of P4P bonuses and low levels of P4P bonuses, also reduce the desired effect of P4P on quality and volume, but they do not cause the outputs to fall below baseline levels. Optimal effect of P4P on quality and volume of services is obtained when P4P bonuses are distributed per the health workers' contributions to the services that triggered the payments. Other distribution algorithms such as equal allocation or allocations proportionate to salaries resulted in quality and volume levels that were substantially lower, sometimes below baseline. The system dynamics models served to inform, with quantitative results, the theory of change underlying P4P intervention. Specific implementation strategies, such as prompt disbursement of adequate levels of performance bonus distributed per health workers' contribution to service, increase the likelihood of P4P success. Poorly designed P4P schemes, such as those without an optimal algorithm for distributing performance bonuses and adequate safeguards for gaming, can have a negative overall impact on health service delivery systems.