A transparent process for creating behavioural systems maps to inform population simulation models which aim to explain and predict public health policy effects: the case of smoking cessation in England

Background: Computer simulations can be used to help inform public health policy decisions by synthesising relevant evidence and comparing the estimated long-term outcomes of existing and new policies. Predicting the impact of policies requires making causal inferences about how the policies affect the outcomes. To make informed predictions, it is necessary to understand and describe how the behaviours of interest are influenced and the policy mechanisms of action upon these influences. However, methods for the development of such simulation models are limited. This work aims to develop a transparent process using behavioural systems maps for creating the structure of a simulation model which can explain and predict public health policy effects, using a case study of smoking cessation. Methods: The PHEM-B toolbox of methods for incorporating the influences on Behaviour into Public Health Economic Models was used as a basis for the methodology. An ‘action-research’ approach was used to develop the process for building the behavioural systems maps, informed by a collaboration between decision modellers and behavioural scientists using the Capability, Opportunity, Motivation-Behaviour (COM-B) model. Results: The process for developing behavioural systems maps for informing policy models involved: (1) identifying the behavioural outcomes to be modelled; (2) linking important socioeconomic, psychological, behavioural, social, economic and structural factors to behaviour using behavioural theory; (3) incorporating mechanisms of action of policies through these factors using evidence; and (4) identifying what is feasible to be modelled. In the smoking case study, a different set of influences were identified affecting probability of attempting to quit smoking and maintaining the quit, leading to different policy choices for these two outcomes. Conclusions: The PHEM-B toolbox and the COM-B framework provided a basis for creating behavioural systems maps to inform simulation models of population behaviour to inform policy choices. The process enabled multidisciplinary discussions about the hypothesised causal relationships, as well as a mechanism through which the model structure could be developed and verified. Using this process to develop other public health models should lead to more informed policy impact predictions.