Large-scale interventions for obesity prevention are rarely well evaluated. Recent concepts of ‘systems change’ as the explicit nature of the intervention have further complicated evaluation designs and measurements. The usual null-hypothesis testing approaches for program evaluation pay attention to the fidelity of a defined intervention implementation, maximizing internal validity, and comparisons with non-intervention groups. If there is an assumption that systems change, rather than program implementation, is the only way to achieve at-scale, sustainable, society-owned outcomes for obesity prevention, the main research question becomes ‘how can existing systems be strengthened or reoriented to reduce obesity prevalence?’ Implicit in this question is a focus on understanding implementation as well as measuring impact, and that interventions are primarily working ‘in’ or ‘on’ existing systems.
Robust measurements of impact on obesity prevalence require comprehensive monitoring systems in place for fine-grained, population data. This occurs in some countries for childhood obesity and is underway in Victoria using opt-out consent approaches. The heterogeneity in monitored prevalence trends can be used to understand systems change through case studies of areas with unexpectedly good or poor trends in obesity.
Measuring implementation and systems change will require the use of other methods such as social network analyses, systems dynamics models, or continuous quality improvement tools. Experience with using these tools for obesity prevention evaluation is in its early days. Hallmarks of systems changes are changes in food and physical activity environments, and there are good methods for measuring these changes. Measuring the mediating changes in behaviours is problematic for theoretical reasons (expected small changes across multiple behaviours) and practical reasons (blunt and burdensome tools) and thus, are a lower priority for evaluation.
In conclusion, the use of measurement tools from other disciplines should greatly enhance the evaluation of systems intervention for obesity prevention.