Methods of Health Behavior Change (2024)

Table 1. Linking Evidence-Based Principles of Behavior Change with Components of the COM-B Model

Table 2. Linking the Theoretical Domains Framework with Evidence-Based Principles of Behavior Change and the COM-B Model

The Behavior Change Wheel is a synthesis of 19 frameworks of behavior change identified in a systematic literature review (Michie, van Stralen, & West, 2011). They include widely used frameworks such as “intervention mapping” (Bartholomew, Parcel, Kok, Gottlieb, & Fernandez, 2011) and “MINDSPACE” (Institute for Government, 2010); the full list and details of the synthesis process can be found in the supplementary files of Michie et al. (2011). These frameworks were evaluated against three criteria that were deemed to be essential characteristics of their usefulness to inform intervention development: conceptual coherence, comprehensive coverage of intervention types, and linkage to a model of behavior. A conceptually coherent framework includes categories of the same type and specificity of entity, and a comprehensive framework offers the full range of potential types of interventions for intervention designers to consider. The third criterion encapsulates the importance for a framework to provide a system whereby one can formally analyze the target behavior based on a theoretical model and systematically select intervention types depending on what influences need to be changed to achieve behavior change. Since none of the frameworks met all three criteria and there was overlap between them, an integrative framework was developed from the 19 frameworks (Michie, van Stralen, & West, 2011).

The Behavior Change Wheel (Figure 1) has the COM-B model at its hub surrounded by two rings of intervention strategies. The first ring has nine intervention functions that are direct methods for changing behavior, with the outer ring being seven policy categories which are high-level societal and organizational strategies for implementing and sustaining behavior change. Each intervention function reflects many different behavior change techniques; these are the smallest intervention components that on their own have the potential to change behavior (Michie et al., 2014; Michie & Johnston, 2013).

The Behavior Change Wheel provides a systematic guide for designing behavior change interventions (Michie et al., 2014; Michie, van Stralen, & West, 2011). It has been used widely across a range of behavioral domains, settings, and problems. Examples include interventions designed to increase engagement with face-to-face smoking cessation support among adults (Fulton, Brown, Kwah, & Wild, 2016), aid smoking cessation among pregnant women (Tombor et al., 2016), improve childhood weight management among parents (Curtis, Lahiri, & Brown, 2015), promote attentive eating among overweight and obese adults (Robinson et al., 2013), reduce cardiovascular disease risk among people with severe mental illness in primary care (Osborn et al., 2016), improve the implementation of upper limb exercise in stroke rehabilitation (Connell, McMahon, Redfern, Watkins, & Eng, 2015) and improve general practitioners’ management of long-term medications (Sinnott et al., 2015).

Selecting effective strategies for behavior change needs to be based on a clear understanding of which behaviors are likely to be most easy to change with greatest impact. This decision should be informed by a “behavioral diagnosis.” A behavioral diagnosis starts with defining the problem that the intervention aims to address (e.g., to reduce childhood obesity) in behavioral terms (e.g., improving children’s diet, increasing their physical activity etc.). For each key candidate behavior, one needs to define the behavior very precisely in terms of who needs to do what differently, when, where, with whom, and how. For example, in order to improve children’s diet, one may suggest that parents need to pack fruits and vegetables in children’s lunchbox every day instead of snacks, or that children need to cut down their daily sugary drink consumption, or that healthcare professionals need to support parents in managing their children’s diet, or that teachers need to include the topic of healthy eating in the school curriculum. Behaviors tackled at policy level include additional categories of people whose behavior is relevant to the target behavior. For example, for food consumption, relevant people would include those involved in the food industry, commissioners of health and social services, town planners, policymakers within local and national government, etc. Given that any behavior is influenced by other behaviors within and between people, both positive and negative “spillover” effects and the significance of these consequences need to be considered.

After specifying the target behavior and population in as much detail as possible (e.g., to help parents pack fruits and vegetables in children’s lunchbox every day instead of snacks), the next step in the behavioral diagnosis involves using the COM-B model or the Theoretical Domains Framework to investigate what needs to be changed in people and/or the environment to bring about behavior change (Francis et al., 2012; Michie et al., 2014; Michie, van Stralen, & West, 2011). A thorough assessment of behavior draws on different sources. This can include evidence from the scientific literature, findings from formal research, and less formal data gathering conducted with people from the target population, stakeholders, and experts in relevant fields by means of interviews, focus groups, and questionnaires (Francis et al., 2012). A behavioral diagnosis contributes to a better understanding of the nature of the target behavior in terms of the intrapersonal (e.g., parents’ knowledge and beliefs about what constitutes a “healthy lunchbox”) and environmental influences on its development (e.g., influences of what other parents in the social environment pack in their children’s lunchbox), maintenance, and predicted levers for future change. This can inform decision-making regarding the interventions and policies that are likely to be effective in a given context (Michie et al., 2014; Michie, van Stralen, & West, 2011).

Within the Behavior Change Wheel, each of the nine intervention functions refers to a general method by which an intervention can change behavior (Michie et al., 2014; Michie, van Stralen, & West, 2011), and these can be linked with evidence-based principles of behavior change (Table 3). The intervention functions are:

For example, a behavioral diagnosis may indicate that increasing parents’ reflective motivation (e.g., their beliefs, intentions, and plans) is one of the important factors to be targeted in interventions aimed at increasing the fruit and vegetable content of children’s lunchbox. One can then draw on the Behavior Change Wheel to identify the intervention functions relevant to changing these: education, persuasion, incentivization, and modeling.

Depending on the intervention functions selected, the Behavior Change Wheel points to seven policy categories to enable or support the interventions (Michie et al., 2014; Michie, van Stralen, & West, 2011). These are:

For example, interventions focusing on modeling to change behavior may be supported by policies involving communication and marketing activities and service provision. Definitions of the intervention functions and policy categories are provided in Michie, van Stralen, & West (2011) and Michie et al. (2014).

Intervention functions can be put into practice by a large number of behavior change techniques; 93 have been identified and organized into a taxonomy in an international project (Michie et al., 2013, 2015). For more information about these techniques, see Michie et al. (2014, p. 145), online training, and a smartphone app that can be found by searching “BCT taxonomy” in the app stores. Intervention designers can use systematic methods to identify which behavior change techniques are likely to be the most appropriate given their target behavior/s, population, setting, and intervention constraints (Michie et al., 2014).

Behavior Change Techniques and Taxonomies

Behavior change interventions are complex, at least in part, because they include many interacting components (Craig et al., 2008). Specifying interventions by behavior change techniques improves the precise reporting of the content of complex interventions so that they can be evaluated, delivered with fidelity, replicated, and generalized to different contexts. Using a shared language when describing intervention content can facilitate this process, and taxonomies of behavior change techniques have been developed with agreed labels and definitions. For example, “action planning” must include detailed planning of performance of the behavior in terms of at least one of context, frequency, duration, or intensity (Michie et al., 2013).

Several taxonomies have been developed for specific behavioral domains including physical activity and healthy eating (Abraham & Michie, 2008; Michie, Ashford, et al., 2011), smoking cessation (Michie, Hyder, Walia, & West, 2011; West, Evans, & Michie, 2011), reducing excessive alcohol consumption (Michie et al., 2012), and increasing condom use (Abraham, Good, Huedo-Medina, Warren, & Johnson, 2012). Building on these taxonomies, the Behavior Change Techniques Taxonomy v1 (BCTTv1) was developed by an international consensus of 400 experts from 12 countries. It includes 93 distinct behavior change techniques that are hierarchically structured and clustered into 16 higher-order groups (Michie et al., 2013, 2015).

An empirical approach, involving open-sort grouping tasks and hierarchical cluster analysis, was used to group behavior change techniques in the BCTTv1 (Michie et al., 2013, 2015). The structure of each higher-order grouping containing similar behavior change techniques in terms of their mechanisms of change can aid easier recall of behavior change techniques and the practical use of the taxonomy. Behavior change techniques in the BCTTv1 can also be reliably linked to domains from the Theoretical Domains Framework. There is a moderate overlap between higher-order groups and theoretical domains. Both of these and the simpler COM-B model can aid the selection of behavior change techniques to inform theory-based intervention development, which is used by researchers appears to reflect personal preference and past practice. Higher-order groups with example behavior change techniques are:

Evidence regarding the effectiveness of behavior change techniques is accumulating for different behaviors and populations, for example, to improve dietary behaviors (Lara et al., 2014), increase sedentary behavior (Gardner, Smith, Lorencatto, Hamer, & Biddle, 2016), and aid physical activity among healthy and overweight adults (Williams & French, 2011), obese adults (Olander et al., 2013), and older adults (French, Olander, Chisholm, & Sharry, 2014). There is some evidence to suggest that interventions that use more behavior change techniques are more effective than those that use fewer behavior change techniques (Gardner et al., 2016; Webb, Joseph, Yardley, & Michie, 2010).

Behavior change techniques may yield greater effects if they are delivered in theoretically informed groups rather than in isolation. An effective combination of behavior change techniques that work synergistically together is self-monitoring with goal-setting, action planning, and/or feedback on behavior (Dombrowski et al., 2012; Michie, Abraham, Whittington, McAteer, & Gupta, 2009). Also, problem-solving has been found to be more effective in combination with providing rewards for behavior (van Genugten, Dusseldorp, Webb, & van Empelen, 2016). A third example is that although using threatening communications about health risks has been advocated to change behavior, such as to aid smoking cessation (Borland et al., 2009), its effectiveness has been found to be greater if it is paired with increasing people’s confidence to change (“self-efficacy”) (Peters, Ruiter, & Kok, 2013).

Relevant research evidence is an important criterion for selecting behavior change techniques for a given behavior, population, and setting. However, there are other criteria to be considered when selecting behavior change techniques that are likely to increase the implementation of interventions.

Systematic Context-Based Decision-Making About Behavior Change Interventions: The APEASE Criteria

The process of intervention development is likely to generate a large number of possible intervention functions, behavior change techniques, and policy categories. However, it may not be possible or appropriate to implement all of these given the context and limitations such as the mode of delivery of the intervention and resources, and thus decisions have to be made as to which methods to include based on explicit reasons (Michie et al., 2014).

The APEASE (Affordability, Practicability, Effectiveness and cost-effectiveness, Acceptability, Safety and side effects, Equity) criteria have been developed to help intervention developers and policymakers follow a systematic decision-making process when designing behavior change interventions (Michie et al., 2014). “Affordability” refers to interventions being designed and delivered within a specified budget and accessed by people from the target population. “Practicability” refers to whether it is feasible to deliver the intervention as intended. “Effectiveness” refers to whether the intervention is effective in relevant real-world contexts. This should be judged based on evidence primarily from high-quality randomized controlled trials and experimental and observational studies. When considering the implementation of effective interventions, more weight should be given to the more cost-effective ones. “Acceptability” refers to whether the intervention is likely to be acceptable to relevant stakeholders, including people from the target population, policymakers, healthcare professionals, funding agencies, and national governments. The criterion of “safety and side effects” refers to the possible unwanted or unintended consequences of an effective intervention. A behavior change intervention meets the criterion of “equity” to the extent that it reduces the disparities between different groups of society and provides equal access to the intervention across the social spectrum.

Although APEASE provides a structured way of assessing intervention methods, intervention designers and policymakers need to draw on local knowledge and their own expertise when selecting the most appropriate intervention functions, behavior change techniques, and policies to be implemented in a given context. The criteria will be illustrated in relation to interventions to increase smoking cessation.

There is ample evidence that the most effective treatments for tobacco dependence are those that combine pharmacotherapy with behavioral support provided by smoking cessation specialist advisors (Kotz, Brown, & West, 2013; Stead & Lancaster, 2012). Although these are both effective and cost-effective interventions (APPG, 2010; West, May, West, Croghan, & McEwen, 2013), some of the more expensive smoking cessation medications (e.g., varenicline) and multi-session counseling may not be globally affordable, especially in low-income countries (West et al., 2015). On the other hand, brief opportunistic smoking cessation advice from healthcare professionals, proactive telephone support, text messaging interventions, printed self-help materials, and some forms of pharmacotherapy (e.g., cytisine) are also effective and appear to be globally affordable to aid cessation (West et al., 2015).

In terms of practicability, effective behavior change techniques, such as providing feedback on smokers’ expired carbon monoxide readings (Shahab, West, & McNeill, 2011), may be considered as more feasible if delivered face-to-face rather than via digital interventions (Tombor, Neale, Shahab, Ruiz, & West, 2015).

Providing financial incentives for smoking cessation in pregnancy has been found to be effective and highly cost-effective, and therefore it may be considered in high-income countries as a valuable addition to existing public health services (Boyd, Briggs, Bauld, Sinclair, & Tappin, 2015; Higgins et al., 2012; Lumley et al., 2009; Tappin et al., 2015).

In terms of acceptability, it has been found that a third of current smokers in England would support a total ban on sale of tobacco products (Shahab & West, 2010). However, if this is not aligned with national governmental objectives, the intervention is unlikely to be supported by relevant policies.

Providing financial incentives is effective in increasing smoking cessation rates among pregnant smokers (Tappin et al., 2015); however, unintended consequences have been identified including creating distrust between healthcare professionals and clients and increasing the negative public perception of pregnant smokers (Thomson et al., 2014). It is important to identify potential negative consequences, and it needs to be considered whether the expected benefits of a behavior change intervention would be likely to outweigh its potential risks.

Smokers from low-socioeconomic groups are less likely to stop smoking than smokers from high socioeconomic groups (Kotz & West, 2009). Even though they are motivated to quit (Kotz & West, 2009), smokers from disadvantaged backgrounds are less likely to attend face-to-face support and use pharmacotherapy for long enough to aid cessation (ASH, 2016; Hisco*ck & Bauld, 2013). Improving equity of access to, and use of, effective smoking cessation support requires interventions to meet the needs of smokers across the social spectrum and target deprived communities where smoking prevalence is the highest (ASH, 2016; Hisco*ck & Bauld, 2013).

The APEASE criteria have been used to inform the development and evaluation of behavior change interventions in various settings, such as to improve antibiotic prescribing in long-term care facilities (Fleming et al., 2014), improve the routine delivery of caries prevention and management in primary dental care (Templeton et al., 2015), optimize general practitioners’ medication management for patients with multiple chronic conditions (Sinnott et al., 2015), select modes of delivery for an intervention targeting multiple health behaviors among students (Tzelepis et al., 2015), and modify an existing evidence-based intervention for sepsis treatment in order to improve its implementation (Steinmo et al., 2016).