Economic evaluation of universal programmes to promote healthy behaviours: challenges and possible solutions with an application to physical activity

Several methodological challenges characterise the economic evaluation of public health interventions, especially universal programmes to promote healthy behaviours. While relevant guidance to support analysts exists, methodological shortcomings have been highlighted in recent reviews. The choice of evaluation method and assumptions can have profound impacts on cost-effectiveness, and consequently decision-making. Therefore, methods development and validation for public health interventions are needed. Narrowing the focus to physical activity, the first research question of this thesis was: if and how have the methodological challenges acknowledged in public health have been addressed in economic evaluations of PA promotion interventions? To address this question, a systematic review of the methods used for economic evaluation was conducted. This review revealed a paucity in the methods used and identified four outstanding issues to be addressed in this thesis: modelling of heterogeneity and inequality in population-level impact, assumptions regarding maintenance of behaviour change over time, longitudinal selection bias and perspective for economic evaluation. To address these four issues, this thesis was framed around two research questions: how to incorporate concerns regarding population-level impacts into the economic model? What is the impact of the key methodological assumptions that underpin the existing models on the economic decision? These questions are explored using a case study which focused on a local universal programme to promote physical activity. A novel modelling approach is proposed to address the issues of population-level impact and behaviour change maintenance over time. Results show the choice of evaluation method can impact the cost-effectiveness decision. In particular, the choice of method and assumptions regarding selection mechanisms and maintenance of behaviour change over time have the potential to independently affect identification of the optimal strategy. Results also suggest ...