There is increasing recognition that interventions to change behaviour should draw on theories of behaviour and behaviour change in their development. For example, in the UK, the Medical Research Council has published a strategy for developing and evaluating complex interventions, which starts with a “theory” phase before progressing to “modelling” and then experimental phases (exploratory trial and randomised controlled trial (RCT)) (Medical Research Council, 2000; Campbell, Murray, Darbyshire, Emery, Farmer, Griffiths, Guthrie, Lester, Wilson, & Kinmonth, 2007). In the theory phase, evidence is accumulated and a theoretical basis for intervention is developed which is modelled in the next phase. Modelling involves hypothesising and testing both what to target (behavioural determinants) and how to do this (techniques to change these determinants). The process of designing and implementing an intervention was seen as challenging: “Problems often arise in the evaluation of complex interventions because researchers have not fully defined and developed the intervention” (Campbell, Fitzpatrick, Haines, Kinmonth, Sandercock, Spiegelhalter, & Tyrer, 2000, p. 694).
There are three main reasons for advocating the use of theory in designing interventions. First, interventions are likely to be more effective if they target causal determinants of behaviour and behaviour change; this requires understanding these causal determinants, i.e. theoretical mechanisms of change. Second, theory can be tested and developed by evaluations of interventions only if those interventions and evaluations are theoretically informed. Third, theory-based interventions facilitate an understanding of what works and thus are a basis for developing better theory across different contexts, populations, and behaviours.
Theory represents an integrated summary of the hypothesised causal processes involved in behaviour change. Unlike “theory-inspired” interventions, theory-based interventions use an explicit causal pathway (Michie & Abraham, 2004) and enable the intervention developer to avoid implicit causal assumptions which may lack evidence or even have been invalidated (Johnston, 1995). Causal processes that underlie a behavioural intervention can be tested within randomised controlled trials examining the effectiveness of the intervention (the Improved Clinical Effectiveness through Behaviour Research Group (ICEBeRG), 2006; Francis, Grimshaw, Zwarenstein, Eccles, Shiller, Godin, Johnston, O’Rourke, Presseau, & Tetro, 2007), thereby strengthening the evidence base for intervention design. Without a theoretical basis, even a large literature on behaviour change interventions may offer no guidance on how to design an intervention for a new situation (Foy, Eccles, Jamtvedt, Young, Grimshaw, & Baker, 2005). In trials of interventions to enhance the implementation of evidence-based practice by health professionals, evidence from over 235 RCTs showed modest success; however, the authors of the systematic review concluded that they had no basis on which to design a new intervention as very few of the trials had used any theoretical foundation and it was therefore impossible to find an integrating framework that could signal the basis of effective interventions (Grimshaw, Thomas, MacLennan, Fraser, Ramsay, Vale, Whitty, Eccles, Matowe, Shirran, Wensing, Dijstra, & Donaldson, 2007).
Nevertheless, even with a theoretical framework, there is little information about how to develop theory-based interventions. A notable exception is Social Cognitive Theory (Bandura, 1997) which specifies how to change the main causal determinant of behaviour, namely self-efficacy, using four techniques: mastery experiences, modelling or vicarious experience, persuasion, and giving physiologically compatible experiences. By contrast, a systematic review of the use of the Theory of Planned Behaviour (Ajzen, 1991) in interventions concluded that the theory was rarely used to design the intervention and was more frequently used as a background to understand the behaviour and to develop measures (Hardeman, Johnston, Johnston, Bonetti, Wareham, & Kinmonth, 2002). Even when people use theory, they tend to use it to explain behaviour but not to change behaviour. For example, Ajzen proposes that the first stage in developing behaviour change interventions is to identify what predicts the behaviour and then to change the predictors, but leaves open the question as to how to change these targets. This is evident in his advice, “Once it has been decided which beliefs the intervention will attempt to change, an effective intervention method must be developed. This is where the investigator's experience and creativity comes into play” (Ajzen, 2006, p. 2). Hardeman, Sutton, Griffin, Johnston, White, Wareham, and Kinmonth (2005) attempt to make the process explicit, but comment that there was no simple link between theory and the choice of intervention techniques.
Thus there is little guidance on how to progress through the early phases of the MRC framework for complex interventions. In considering the key tasks in optimising an intervention, Campbell et al. (2007) do not even refer to the theory-base identified in conceptualising the target problem. Hardeman et al. (2005) have proposed a causal modelling approach (see Box 1). Each arrow represents a causal process and interventions are targeted at changing these causal processes. Within this framework, behaviour change is achieved by targeting the determinants of behaviour.
Behavioural determinants (step 1) can be identified from theories of behaviour. So for example, the Theory of Planned Behaviour (Ajzen, 1991), Social Cognitive Theory (Bandura, 1997), and Operant Learning Theory (Skinner, 1963) all propose, and have evidence from cross-sectional and longitudinal studies to support a range of constructs which affect behaviour (Walker, Grimshaw, Johnston, Pitts, Steen, & Eccles, 2003) including: intention, perceived behavioural control, self-efficacy, outcome expectancy, response-reinforcement contingencies, and discriminative stimuli. Evidence-based behaviour change techniques can be directed at these identified behavioural determinants, and evidence for their role in behaviour change investigated (Michie, Hardeman, Fanshawe, Prevost, Taylor, & Kinmonth, 2007). However, effective mapping of theoretical constructs to behaviour change techniques also requires work to: (1) address the problem of the wide range of theoretical frameworks available; (2) specify the range of techniques available to change the determinants of behaviour; (3) develop a basis for selecting relevant techniques to map on to differing determinants of behaviour.
Ideally, researchers designing interventions would choose a small number of the theoretical frameworks based on empirical evidence of their predictive and intervention value, i.e. there should be evidence that the theory can predict the behaviour and that interventions which change these determinants achieve change in behaviour. However, where that is lacking, it would be valuable to find a systematic way to simplify these potential determinants. Two independent attempts at simplification, based on expert consensus, have been published (Fishbein, Triandis, Kanfer, Becker, Middlestadt, & Eichler, 2001; Michie, Johnston, Abraham, Lawton, Parker, & Walker, 2005a) and show good agreement about the key behavioural determinants (see Table 1).
|Fishbein, Triandis, Kanfer et al., 2001||Michie, Johnston, Abraham et al., 2004|
|Self-standards||Social/professional role and identity|
|Self-efficacy||Beliefs about capabilities|
|Anticipated outcomes/Attitude||Beliefs about consequences|
|Intention||Motivation and goals|
|Memory, attention, and decision processes|
|Environmental constraints||Environmental context and resources|
However, there is still the need to identify techniques to change these behavioural determinants, as illustrated in Box 1. The work reported in this paper relates to the development of methods to clarify the list of behaviour change techniques (step 2) and to identify links between the behaviour change techniques and behavioural determinants (step 3).
There is a wide range of techniques available to change behavioural determinants. These techniques are described in texts largely used by applied psychologists and mental health practitioners. However, they tend to be integrated with other techniques designed to change mental states rather than behaviour and are presented as practical tools without reference to their evidence base or clear indication of which theoretical constructs they might target. There is currently no comprehensive and accessible list of techniques; it would be extremely difficult for someone new to the field of behaviour change to extract these techniques from the literature and to find the detail that would be necessary to use them in a complex intervention. In addition, their appropriate application depends on mapping these techniques on to the proposed behavioural determinants.
Thus this paper reports the development of a procedure for selecting relevant techniques to map on to each of the behavioural determinants. It seems obvious that different techniques will address different behavioural determinants. For example, it might be appropriate to rehearse practical skills where the determinant is lack of skill, but not where there is lack of motivation to perform the skill. This mapping process is essential if we are to optimise the benefits of theory-based interventions. Other approaches to intervention development have not done this work, e.g. MRC framework, Intervention Mapping (Kok, Schaalma, Ruiter, Van Empelen, & Brug, 2004).
This paper describes two pieces of preliminary work addressing steps 2 and 3 (Box 1). The first was to develop an extensive list of behaviour change techniques and definitions; the second identified links between these techniques and the theory-based behavioural determinants identified through step 1 and specified in Table 1. This paper describes a first iteration of this process and provides a basis for the further elaboration of this work (dealing with issues such as the further identification of techniques, completing technique definitions, and the elimination of overlap between techniques). Our aim is to contribute to a process of constructing an evolving taxonomy of behaviour change techniques to be used for developing theory-based behaviour change interventions.