Developing methodology for designing and evaluating theory-based complex interventions: an ontology for linking behaviour change techniques to theory

Behaviours (such as smoking, excessive alcohol use, being sedentary and eating unhealthy foods) result in serious illness and early death. Efforts to change such behaviours are a priority in most health systems. Interventions have been developed that can help support changes in these behaviours. However, these interventions are made up of many components and we often do not know which components are responsible for bringing about change nor how this change, which is often quite limited, occurs. This restricts our ability to learn from our research to inform the development of more effective interventions. There is a gap between our empirical knowledge of which techniques work and our understanding of why they work. At the present time our understanding of the mechanisms is derived from various theories which are not linked systematically either to the interventions or the mechanisms - there are many techniques and many theories - but how they link together is far from clear. The aim of this project is to establish precisely the links between behaviour change techniques and their mechanisms of action. These links provide the basis for a framework that can be updated over time and which will permit the development of more advanced intervention design and evaluation, thus speeding up the process of building our knowledge about 'what works'.

At present, individual experts make judgments about links between behaviour change techniques and theoretical mechanisms, without a systematic, shared or agreed method of doing so. We therefore plan to work with a large number of international experts from many disciplines to develop a method for reaching a consensus on articulating the theoretical links between techniques and mechanisms. We will also examine the literature to investigate how behaviour change techniques have been linked with theory in published studies of behaviour change interventions. Experts will consider the combined findings of the consensus and literature-based studies to produce a large matrix of links between behaviour change techniques and theoretical mechanisms (a behaviour change 'ontology'). We will disseminate our findings so that they are reported and discussed widely and scrutinised by the international scientific community.

We will provide web-based resources and give workshops to introduce this new tool, the behaviour change (BC) ontology. We believe this work will provide a more robust method for researchers to use in developing soundly based interventions and in interpreting the findings of interventions designed by other researchers, resulting in more effective interventions and health benefits.

Despite the recommendations that complex interventions to change behaviour should be informed by theory, there is a lack of guidance as to how to do this. We need a reliable, transparent method to link behaviour change techniques (BCTs) to theoretical mechanisms (operationalised as 'constructs' or 'domains'). This linkage ('ontology') will enable researchers to apply a systematic, shared method for (i) Intervention design: selecting BCTs on the basis of their judged mechanism/s of action, (ii) Evidence synthesis: aggregating interventions by BCT and/or hypothesised mechanism and (iii) Understanding mechanisms of action in order to both optimise interventions and develop theory. Such a behaviour change (BC) ontology requires the development of a method to produce it.

This project aims to:
1. develop a systematic and replicable method for generating a BC ontology of reliably identified links between BCTs and theory (operationalised as theoretical domains and component constructs) (Studies 1-4).
2. produce a web resource and manual including guidance on how to use the ontology in intervention design and evaluation, and evidence synthesis, illustrated with examples from effective interventions (Activity 1).
3. maximise the engagement of the international scientific community in the use and future development of the BC ontology (Activity 2).

Our preliminary consensus method for linking BCTs and theoretical domains (Michie et al, 2008) has proved popular with intervention designers wishing to draw on behaviour change theory to inform interventions, an obvious group of users for the proposed methodology and methodological tools we seek to develop. For example, an intervention designer will be able to select from identified groups of BCTs that are theoretically proposed to work together, using full information about the strengths of these links and possible sources of weakness. This research will give intervention designers useable tools to consider what mechanisms of action are likely to be most relevant for their situation and, therefore, which BCTs are likely to be most effective.

Systematic reviewers will be able to link BCTs identified in intervention descriptions to the most plausible theories and evaluate empirical support for theoretical constructs/domains as a basis for designing effective future interventions. Creating a BC ontology will allow clearer thinking and a more systematic method, leading to the more rapid development of effective interventions. It will also provide computer scientists and those in behavioural informatics a tool for interrogating the published literature and 'Big Data' to efficiently build an evidence base of which BCTs have what effects with what mechanisms of action. Three research groups in the US have been in touch with SM asking about plan to produce a BC ontology that they could use for this purpose.

More effective interventions will benefit the health of the general population. They will additionally aid patients in achieving a wide variety of behaviours to help them manage their condition and symptoms. Finally, they will improve health care and the delivery of public health by increasing the extent to which practitioners follow evidence-based guidance and by strengthening evidence-based health policy.