Health economic modelling of public health interventions
Public health interventions may include actions or activities that aim to make a person or population change behaviours to improve their health. They are often multi-component and operate within complex systems, which means that there is not a clear boundary around the system and the sum of individual intervention effects is not equal to the outcomes at the population level due to the interactions between heterogeneous individuals and influence of their environment. Thus, effects of public health interventions may be non-linear, and sometimes unexpected, at the macro level, with the wider context impacting intervention effectiveness (Skivington et al., 2021).
Health economic models are used to predict the difference in costs and effectiveness between current practice and alternative interventions, usually over an individual’s lifetime, to capture all impacts of the interventions to inform policy decisions about how best to spend limited resources (Briggs et al., 2006; NICE, 2023). The benefit of these models is that they can synthesise evidence from a range of sources and simulate possible future costs and outcomes of alternative interventions. Existing health economic models do not typically incorporate the determinants of individual and population behaviours that influence long term effectiveness, yet it is essential to understand how public health interventions work in order to attempt to predict long-term outcomes of interventions.
Health economic models include simple arithmetic calculations, cohort state transition models, and individual level simulations (Brennan et al., 2006). More flexible individual-level health economic models can be useful when decision makers want to understand the different impacts of interventions upon individuals or different groups of the population to reduce health inequalities (Boyd et al., 2022) or when outcomes depend on time or event-dependent interactions (Emmert-Fees et al., 2021). Typically, within these models, a population is synthesised to match the characteristics of the real population of interest, so that every individual has their own attributes (e.g. age, socioeconomic status, Body Mass Index (BMI)), which can be updated over time (Wu et al., 2022). These models can then be used to estimate the incidence and progression of diseases and health conditions using epidemiological risk equations, to estimate mortality, to assign different resource use, costs and utilities, to test the impact of alternative interventions, and to report outcomes by relevant subgroups.
Studies assessing the effectiveness of public health interventions, like smoking cessation or exercise classes, often report at 6 or 12 months. To inform policy decisions, it is useful to predict the relative costs and benefits of these interventions over the longer term. Health economic models can be used to do this, but they usually make simple assumptions about the impact of interventions on longer term behaviour based on limited evidence or theory, and these assumptions can substantially affect results. To date, there has been little research about how the influences on behaviour can be incorporated in health economic models.
Aim of this toolbox
This toolbox aims to draw on psychology, sociology, behavioural economics, complexity science and health economics to help modellers, behavioural scientists and policy makers to identify and incorporate appropriate methods and behavioural theory within health economic models of public health interventions, to better inform resource allocation decisions.
For each method, a description, an expert consensus on the circumstances when they could be used, how the method can bring behavioural influences into the model, the minimum resource requirements and key references are provided.
However, there are weaknesses in current behavioural theories, methods and the data available to inform such models. Thus, an agenda for further research for both health economic modellers and behavioural scientists is also set out.