3. Application of ontology 

The behaviour change intervention ontology (BCIO) has been developed to provide consistent language within which to describe interventions, including the mode of delivery, setting, source, schedule and dose, population, method of engagement, style of delivery, behaviour change techniques (the content of the interventions), mechanisms of action (how the interventions work), human behaviour and intervention fidelity (Michie et al., 2020). The TIDieR Checklist has previously been developed to improve reporting of interventions (Hoffman et al., 2014), which includes many of these elements and may be less resource- intensive. However, the benefit of the BCIO is that for each label in the ontology there is a unique ID number. This can facilitate consistency between intervention development and evaluation, and across models and theories. It can also enable the synthesis of similar interventions within meta-analysis which can be computer automated, which can be used within the health economic modelling. Specifying the detail of the interventions in this way could also help in informing the behavioural systems map and understanding the potential long-term impacts of the interventions. It is yet to be determined whether the BCIO fully captures all elements of interventions, for example behavioural economic interventions; however, the intention is that the ontology will be updated accordingly within future versions when elements are found to be missing.  

Where a set of interventions are being assessed in combination in the model, but effectiveness evidence is only available for the individual interventions, the content of the interventions and behavioural theory could be used to understand the mechanisms of action. If they operate through completely different mechanisms because their content is different, then assuming an additive effect may be appropriate; and if some overlap, then a multiplicative effect. In addition, based upon the themes identified for behaviour maintenance (Kwasnicka et al., 2016), maintenance motivation, resources, environmental support and self-efficacy are all required for behaviour maintenance. Thus, interventions which impact all of these, when combined, may substantially increase the probability of behaviour maintenance compared with any of the interventions alone. Since within a health economic model it is necessary to predict the impact of the interventions over the long term, this could be used to help choose plausible long-term assumptions of intervention combinations.