2
Ogden, J. (in press). Celebrating variability and a call to limit systematisation: the example of the Behaviour Change Technique Taxonomy and the Behaviour Change Wheel. Health Psychology Review
Celebrating variability and a call to limit systematisation: the example of the Behaviour Change Technique Taxonomy and the Behaviour Change Wheel
Jane Ogden
Professor in Health Psychology
School of Psychology, University of Surrey, UK
Corresponding author:
Jane Ogden PhD
School of Psychology
University of Surrey
Guildford GU2 7XH, UK.
Tel: 01483 686929
email:
Celebrating variability and a call to limit systematisation: the example of the Behaviour Change Technique Taxonomy and the Behaviour Change Wheel
Within any discipline there is always a degree of variability. For medicine it takes the form of Health Professional’s behaviour, for education it’s the style and content of the classroom and for health psychology it can be found in patient’s behaviour, our theories and our practice. Over recent years attempts have been made to reduce this variability through the use of the Behaviour Change Technique Taxonomy, the COM-B and the Behaviour Change Wheel. This paper argues that although the call for better descriptions of what we do are useful for clarity and replication this systematisation may be neither feasible nor desirable. In particular, it is suggested that the gaps inherent in the translational process from protocol to behaviour will limit the effectiveness of reducing patient variability, that theory variability is necessary for the health and well being of a discipline and that practice variability is central to the professional status of our practitioners. It is therefore argued that we should celebrate rather than remove this variability in order for our discipline to thrive and for us to remain as professionals rather than technicians.
In the early 1980’s research in primary care identified the problem of doctor variability with doctor’s offering different advice to different patients for a range of problems including asthma, diabetes and hypertension (See Marteau and Johnston, 1990 for a review). Researchers recorded, coded and measured doctor patient interactions to identify the cause of this problem and a solution emerged in the form of evidence based medicine, clinical guidelines, decision trees and financial incentives to ‘encourage’ GPs to do the right thing (Stiles, 1978; Moscovitz, Kuipers and Kassiner, 1988; Roter et al 1997; Sackett, 1995; Roland, 2004). Thirty years later primary care is in crisis as recruitment is poor and GPs either leave clinical work to take up teaching or management or retire early as they struggle with burnout often citing that they have been de-professionalised and that the art has been taken out of their science (Saunders, 2000; BMA, 2015; Schimpff, 2015).
Now in health psychology we have our own problem to solve – that of patient variability. Much of our work is concerned with promoting healthy behaviours yet however much we try, people continue to eat poor diets, obesity is still on the increase, many smoke, most won’t do the recommended level of exercise and adherence to medication and screening is often poor (eg. Mokdad et al, 2004). And so we have developed our own solutions. First we offered patients knowledge through education but soon this was found to be lacking as their beliefs loomed large and it became obvious that knowing did not translate into doing (see Conner and Norman, 2015 for a review). Then we created social cognition models with their emphasis on beliefs as predictors of patient behaviour but quickly recognised that there was a problematic gap between what people intended to do and what they actually did and research highlighted the role of affect, social support, coping and all sorts of chance events that could knock an intention off its path (Orbell and Sheeran, 1998; Sheeran, 2002; Webb and Sheeran, 2006). Next we created interventions to plug this gap such as self affirmation tasks, planning or implementation intentions (Gollwitzer, 1993; Gollwitzer and Sheeran, 2006; Steele, 1988; Epton et al, 2015). These worked to an extent but not for all the people all of the time and so we recognised that we had many more interventions to offer if we drew upon our theoretical roots of behavioural and cognitive psychology. And so we linked in with the world of clinical psychology and complex interventions became our new tool. Trials were run, interventions offered and the mode of intervention was varied to include simple leaflets, face to face interventions and new technologies such as apps or web based approaches. And papers were published which unfortunately showed that yet again we can only change behaviour for some of the people for some of the time (and not for very long) (eg. NICE, 2013; Bull et al, 2014). The problem of patient variability was proving to be a tricky one.
But in 2008 there was a new kid on the block in the form of the Behaviour Change Technique Taxonomy (Abraham and Michie, 2008; Michie et al, 2013). This has burgeoned over the past 7 years into a multi site, multi disciplinary and hugely successful enterprise aiming to code protocols, to explore which techniques are used in which interventions, to identify which techniques are most successful for which behaviours, to train health professionals to select and use the most effective techniques and to provide a resource for all to make behaviour change interventions effective (Atkins, Wood and Michie, 2015; Michie and Wood, 2015). The mission was twofold. First it aimed to describe and systematise the techniques used to change behaviour so that protocols could be clearer and studies could be replicated (eg. Michie and Abraham, 2008; Michie et al, 2011a). A useful and constructive contribution to the discipline. Second it aimed to use this taxonomy to identify which techniques are most effective as a means to maximise the effectiveness of our interventions (eg. Michie et al, 2009; 2011b; Michie and Wood, 2015). Our interventions will be perfected, patients will change their behaviour and the problem of patient variability will be eradicated at last. But will it? If there is a gap between a person’s own beliefs and their own subsequent behaviour (and we know there is) then there is certainly a gap between their own behaviour and that of the health professional (or intervention) in front of them. There is also a gap between that health professional’s own beliefs (‘I intend to practice in line with my training’) and their behaviour with the patient. Further, there is yet another gap between the health professional’s beliefs and the training they received to deliver the intervention, not to forget the gap between the protocol and how it is translated into training or the final gap between the coder of the protocol and the protocol. This is an awful lot of gaps between the protocol being coded and the behaviour of the target of that protocol – the patient. If their own beliefs don’t predict what patients do next it seems unlikely that what was in the protocol telling the professional what to do with them will do either. Probability theory teaches us that effect sizes diminish as they are translated down a chain of action. And psychology research clearly shows that people behave in response to whole array of intra personal, inter personal and external factors rather than just what is done to them. The mission of the BCT is admirable and its desire for clarification and replication has to be a good one. But it seems that there are too many gaps in the system for what is in a protocol to have much to do with what a patient actually does and that we will be left with patient variability for a while longer yet.
But patient variability is not the only problem facing health psychology. We also have the challenge of theory variability. Over the past 50 years psychology, and more recently health psychology, has developed, tested and applied a range of theories including broad perspectives such as behaviourism and social learning theory to more specific perspectives (or models) such as the Theory of Planned Behaviour, Health Belief Model, Self Regulation Theory, Stages of Change, PRIME theory (see Conner and Norman, 2015 for a review) Furthermore, researchers have drawn upon these theories and (models) to develop a number of frameworks of behaviour change including more academic perspectives such as intervention mapping, MINDSPACE and STD / HIV framework as well as those used by the private sector such as Weight Watchers and Slimming World which similarly draw upon psychological theory (see Michie, van Stralen and West, 2011 for a review; NICE, 2013). At times this number of theories (models and frameworks) can seem unnecessary and confusing and several researchers have identified degrees of similarity between them calling for theory integration (Hagger, 2009). For example, Lippke and Plotnikoff (2009) called to integrate the Protection Motivation Theory with the Stages of Change model; Hagger and Chatzisarantis (2009) mapped out a means to integrate the TPB and Self Determination Theory and Gibbons, Houlihan and Gerrard (2009) proposed the prototype- willingness model as an integration of intentional social cognitive with dual process perspectives. One integrated model which has transformed research over recent years is the COM-B with its focus on capability, opportunity and motivation and the associated Behaviour Change Wheel (Michie, van Stralen and West, 2011; Michie et al 2014; Michie and Wood, 2015). This emerged out of an analysis of 83 theories and 1659 constructs by a cross disciplinary team of researchers in terms of three dimensions: comprehensiveness, coherence and a clear link to an overarching model of behaviour. It has since has been proposed as a new integrated framework for behaviour change across a number of domains including physical activity, weight loss, hand hygiene, dental hygiene, diet, smoking, medication adherence, prescribing behaviours, condom use and female genital mutilation (eg. Jackson et al, 2014; Brown et al, 2015; Bailey et al, 2015; Chadwick and Benelam, 2013; Asimakopoulou and Newton, 2015; see Michie and Wood, 2015; Atkins, Wood and Michie, 2015 and Michie et al, 2014 for reviews).
The goal has therefore been to reduce theory variability and identify an integrated systematised approach which transcends individual perspectives and can be applied to all behaviours. Such integration is admirable in line with the aims of ‘eliminating gaps in theories, reducing redundancy and increasing parsimony’ (Hagger, 2009). It also reflects the need to ‘present a ‘streamlined’ and ‘accurate’ view of the processes that underpin health behaviour’ (Hagger, 2009). But does such streamlining and integration help a discipline? Is it good for research? And does it promote and facilitate creativity in those that do the thinking? In his classic text ‘The Structure of Scientific Revolutions’ Kuhn (1962) described the processes involved in ‘normal science’ and argued that whilst those working within any discipline at any time may believe that they are practicing ‘problem solving’ they are actually engaged in ‘puzzle solving’ as the answers have already been determined by the paradigm within which they work. In line with this, the empirical research of those working within the social studies of science illustrated how laboratory scientists practice and perpetuate ‘normal science’ which is restricted in its vision due to the limitations imposed by the paradigm (eg. Latour, 1987; Woolgar, 1988; Mulkay and Gilbert, 1982). Further, they illustrated how through the strategies of serial publications, authorship, conference presentations, collaborations, jargon, networks and rhetoric the dominant ideas become ‘black boxed’ and accepted as truths as they move beyond debate or critique. In addition, Latour (1987) argued that researchers practice ‘action at a distance’ as leaders in a discipline employ and activate their disciples to carry out research to support and perpetuate those ideas dominant in a discipline at any given time. Accordingly, new ideas which challenge dominant views, data which doesn’t fit, and even research groups who try to present opposing perspectives are marginalised as the black boxed constructs become the only acceptable way in which any ‘problem’ can be solved (Latour, 1987; Ogden, 2002). In line with this, the recent move within health psychology to deal with the problem of theory variability by streamlining and integrating our existing perspectives into one dominant model may help remove overlap and redundancy. But at the same time this drive may constrain and limit the discipline in a such a way that we become simple ‘puzzle solvers’ not ‘problem solvers’ , that our science becomes ‘normal’ not novel, ideas become ‘black boxed’ and beyond challenge and any creative anomalies are marginalised before they have the chance to reach fruition. Science requires paradigm shifts if it is to develop and grow. And scientific revolutions need to occur if scientists are to remain free, independent and creative. Removing the variability in our theories may remove the mess in the system. But this very mess may be what keeps a discipline alive.
Health psychology has therefore struggled with variability in both patient behaviour and theory and recently a systematising approach has been developed for both these problems in the hope that once measured, coded, assessed and integrated these problems will be solved. But what about variability in practice? Practitioners are trained to have many different skills and many varied tools in their tool kits. They are also trained to selectively draw upon these tools given the needs of any particular patient at any particular time and are therefore inherently flexible in line with notions of tailored interventions and patient centredness (Kreuter et al, 1999; Mead and Bower, 2000; Richard, Coulter and Wicks, 2015). Further, clinical research frequently shows that the most effective component of any clinical interaction is the therapeutic or working alliance between client and therapist not what the therapist ‘does’ (Godfrey et al, 2009; Bordin, 1975; Ardito and Rabellino, 2011; Cook et al, 2015). Yet this variability is also now being ‘solved’ and not only have the BCT and COM-B become the means to systematise patient behaviour and theory they are also becoming a force to systematise practice which is particularly apparent in a series of papers specifying which techniques should be used for which behaviours (Michie et al, 2009; 2011b), the comprehensive guide published in 2014 (Michie et al, 2014) and its associated websites. The goal to promote evidence based practice has to be a good one; as does the mission to describe what should and is being done so that studies can be evaluated and replications can be carried out. But the goal to identify intervention tools that should be used for a specific behaviour ignores the need for flexibility, variability and change according not to the type of behaviour, or the type intervention or even the type of patient but how that individual patient happens to feel, think, look, behave or respond at any particular time.