Modelling the impact of physical activity on public health: A review and critique

Background: While several reviews have assessed economic evaluations of physical activity in public health and, in most cases, found the interventions to be cost-effective, the validity of the conclusions reached depends on the appropriateness of the modelling methods used in the individual studies. Objective: To provide an overview and critique of modelling approaches and key structural assumptions used in applied studies to estimate the impact of physical activity on population health. Methods: Electronic databases were systematically searched for relevant model-based economic evaluations. A thematic approach was used to assess the modelling studies. The critique determined the appropriateness of the modelling frameworks and plausibility of key structural assumptions. Results: Twenty-five models were identified. Cohort models were most frequently used. High variability in the modelling of downstream diseases was found across studies analysing similar populations. Structural assumptions regarding the dynamics of change of physical activity were unrealistic in most cases. Heterogeneity was addressed in only a few studies, while health equity concerns were, at best, acknowledged by authors. Conclusions: This literature is predominantly characterised by modelling approaches that may not adequately address the complexities associated with representing the physical activity behaviour- population health process. A consensus on how to model the impact of physical activity on public health and development of a reference model could help reduce these sources of uncertainty.