Agenda for further research

Currently, within most policy making arenas there is insufficient time and resources allocated to evaluating the cost-effectiveness of public health interventions, leading to very simple models of these complex systems being developed. Meanwhile, the studies of the interventions are very short term, not always clearly described, and with aggregate results presented. Therefore, there is a substantial further research agenda across fields to advance methods for incorporating the influences of behaviour into health economic models in order to better inform public health policy.

1)   Develop collaborations between health economic modellers and behavioural/ social scientists to inform intervention development, to help understand at an earlier stage whether interventions are likely to be cost-effective and to ensure that useful outcomes for the health economic modelling are collected and reported. A process for working together effectively could be developed following the use of the toolbox within case studies. Health economic modellers could work with behavioural/ social scientists to understand how behaviour maintenance theories might best be utilised within health economic models, and what further research would be beneficial to improve long-term predictions of intervention effectiveness. The authors plan to set up a new network between modellers and behavioural scientists to encourage collaboration and to share resources.

2)   Develop a consensus statement on the most appropriate behavioural theories for each health-related behaviour, ideally through collaboration between psychologists, sociologists and behavioural economists. Subsequently, develop and collect relevant standardised measures of behaviour and influences on behaviour, which use a consistent ontology. Collect longer term data where possible when evaluating the effectiveness of interventions. This could be done by using mobile phone apps or wearable sensors. Develop and test behaviour maintenance theories for different health-related behaviours.

3)   Develop a suite of public health economic agent-based models which are built flexibly and reported open source, including coding, which would allow model reuse and adaptation. This would allow modellers who have limited resources and time within the decision-making process to build upon existing models. Standard social network structures and GIS data could be included which can be modified if both feasible and necessary. If these ABMs were built consistently across model behaviours, then they could link together if behaviours affect each other. Collaboration between health economic modellers and software engineers could improve model development efficiency and reuse. Exploiting recent advances in artificial intelligence may also facilitate this. Evaluate the benefits of the ABMs over standard health economic modelling approaches.

4)   Develop methods for informing long term assumptions about intervention effectiveness. Test the appropriateness of developing expert panels and applying elicitation approaches to help inform structural assumptions and quantify parameters where there are no data. This could include lay people with relevant lived experience. Assess the feasibility of combining qualitative analysis with health economic modelling to inform behavioural assumptions. Explore the potential of utilising GP records (NHS digital) to assess the long-term effectiveness of interventions. Evaluate the benefits of these approaches.

5)   Train modellers to utilise the new methods via short courses, webcasts, and workshops which would need to include an overview of the rationale and the methods, as well as demonstrating the use and outcomes with an example. Within the training, modellers could be given an opportunity to practice the methods using a simple example.