Chapter 2: Literature review
2.7 Intervention development
2.7.3 Theoretical frameworks for intervention development
Intervention development frameworks provide guidance to researchers in a step-wise manner to translate behaviour change theory into practice. O’Cathain et al. (2019) identified six such frameworks within the theory and evidence-based approach, namely the MRC framework for developing and
41 evaluating complex interventions, the Behaviour Change Wheel (BCW), Intervention Mapping (IM), Matrix Assisting Practitioner’s Intervention Planning Tool (MAP-IT), Normalisation Process Theory (NPT) and finally the Theoretical Domains Framework (TDF). A summary of each is provided below which includes strengths and limitations as identified by the designers themselves, other designers, as well as the MRC overview team in the INDEX study (IdentifyiNg and assessing different approaches to DEveloping compleX interventions).
A. UK MRC framework for developing and evaluating complex interventions
The 2008 UK MRC framework is an adaptation of the 2000 version after several papers identified limitations in it. Some of these included that greater detail regarding early phase piloting and development was needed [Hardeman et al., 2005], that a model of the evaluation process should be less linear [Campbell et al., 2007a], that the process and outcome evaluation should be more integrated [Oakley et al., 2006] and that tailored interventions are more successful than standardised ones [Campbell et al., 2007b]. A further updated UK MRC framework has since been released as considerable developments have occurred in the field of complex intervention research [Skivington et al., 2021]. The new framework identifies four main phases, along with key functions and activities that form part of each phase. Skivington et al (2021) state that the phases are often neither linear or cyclical in sequence, and that designers tend to go back and forth between the phases [Skivington et al., 2021]. The four phases as outlined by the MRC are as follows:
Phase 1: Developing or identifying a complex intervention
Researchers are encouraged to use the comprehensive guidance provided by O’Cathain et al. (2019) when developing an intervention. The steps include adequate planning, involving stakeholders, assembling an intervention development team with relevant expertise, reviewing existing literature and theories, testing and refining programme theory, drawing on primary data, making an effort to understand the context of the intervention, considering future implementation, refining the intervention as appropriate and ending the development phase [Skivington et al., 2021].
When using an existing intervention, it is important that the theoretical basis of the intervention or the programme theory (an understanding of how the intervention produced an intended or actual outcomes) is clearly identified, as researchers may not have been included during the development phase [Skivington et al., 2021].
Phase 2: Feasibility testing
Feasibility studies should address any uncertainties identified during the development of the intervention. Progression criteria should be used to guide the decision on whether or not to proceed
42 to the next phase and can include but are not limited to recruitment and retention rates, capacity to achieve intended sample size, feasibility of evaluation outcomes, feasibility to collect and analyse data, consideration of any unintended outcomes, acceptable intervention content and delivery and adherence. Qualitative and quantitative data may therefore be required for this phase, and more than one study may be needed before a full-scale intervention is implemented [Skivington et al., 2021].
Please refer to section 2.9 for further detail on feasibility testing.
Phase 3: Full scale evaluation
Full-scale evaluation involves a tailored approach which should be followed when selecting a study design. As there are a number of available study designs to select from, it is important that this decision is based on the proposed research question, which in turn should be determined by the identified uncertainties surrounding the intervention. The effectiveness alone of an intervention is not the only consideration when making this decision and other factors such as context and budget should also be evaluated [Skivington et al., 2021].
Step 4: Implementation
Implementation science is focused on ensuring that interventions which have already demonstrated effectiveness can be effectively implemented, maximising the impact of the intervention. The main outcomes in an implementation study can include reach, policy impact and utilization of a service [Skivington et al., 2021].
Strengths of the UK MRC framework
The MRC does not regard its framework as prescriptive but rather as a guide for researchers to make the appropriate choices regarding methodology and practicality, similar to the previous frameworks [Skivington et al., 2021; Craig et al., 2008]. Researchers consider the MRC framework as highly influential as it has been widely cited and used by researchers in primary research, as well as in grant proposals [INDEX, 2019; Craig & Petticrew, 2013; Campbell et al., 2000]. The previous MRC framework was found to encourage a practical, tailored approach when selecting evaluation methods, using randomised controlled trials vs observational methods, the use of qualitative and/or quantitative data and outlining the criteria upon which the decision is made [Craig & Petticrew, 2013]. This remains true for the 2021 framework.
The 2008 framework was criticised for not providing great detail regarding its use, which can make the approach difficult to incorporate in the development of interventions [Wight et al., 2016; French et al., 2012]. This has been attended to in the 2021 framework where case studies are presented,
43 demonstrating a variety of approaches. In addition, a checklist has been included to support and document application of the framework, with particular attention paid to the core elements and important considerations for each phase in the research development process [Skivington et al., 2021].
Limitations of the UK MRC framework
The developers of the framework, acknowledge that there were still many issues that were being debated regarding the evaluation of complex interventions at the time of the release in 2008 as well as in 2021 [Skivington et al., 2021; Craig et al., 2008]. Their recommendations for the future include that the selected perspectives and approaches should be those that are more likely to produce a clear and unbiased estimate and should not be limited to those that are less risky to commission [Skivington et al., 2021]. The 2008 framework was also found to be limited in terms of its ability to grasp the real complexity of interventions in health services and the settings in which they are implemented [Craig
& Petticrew, 2013]. It appears that the researchers tried to address this issue in the revised framework, which can only be critiqued once the implementation of the new framework in intervention development is analysed.
B. Intervention mapping (IM)
IM is a six-step intervention planning approach which guides intervention designers from identifying a need or problem to finding a solution by integrating theory and evidence [Bartholomew et al., 2016].
Its designers, Bartholomew et al. (2016), do not consider the steps of IM as linear, but rather that designers move back and forth between tasks and steps. The product of each step informs and guides the subsequent step. The six steps of IM as outlined by Bartholomew et al. (2016) are outlined in Table 2.4.
Table 2.4: Description of the six steps and tasks associated with Intervention Mapping
Steps Tasks
Step 1: Logic model of the problem Establish and work with a planning group
Conduct a needs assessment to create a logic model of the problem
Describe the context for the intervention, including the population, setting and community
State program goals Step 2: Program outcomes and objective – Logic model of
change
State expected outcomes for behaviour and environment Specify performance objectives for behavioural and environmental outcomes
44
Select determinants for behavioural and environmental outcomes
Construct matrices of change Create a logic model of change
Step 3: Program design Generate program themes, components, scope and
sequence;
Choose theory and evidence-based change methods;
Select or design practical applications to deliver change methods
Step 4: Program production Refine program structure and organisation Prepare plans for program materials Draft messages, materials and protocols Pre-test, refine and produce materials Step 5: Program implementation plan Identify potential program users
State outcomes and performance objectives for program use
Construct matrices of change objectives for program use Design implementation interventions
Step 6: Evaluation plan Write effect and process evaluation questions;
Develop indicators and measures for assessment;
Specify the evaluation design;
Complete the evaluation plan
C. Matrix assisting practitioner’s intervention planning tool (MAP-IT)
In 2017, MAP-IT was developed by Hansen et al. by integrating three existing frameworks resulting in a tool which combines research to practice and thereby assisting practitioners to design complex interventions. It is theory-driven, evidence-based, identifies environmental and individual mechanisms and links these mechanisms to behaviour change techniques. The frameworks integrated into MAP- it include the BCW (see section F for details), IM (see section B for details) and Analysis Grid for Environments Linked to Obesity (ANGELO). Hansen et al. (2017) found these frameworks to be helpful but also challenging to work with, time-consuming and extensive knowledge regarding motivational and voluntary processes were required. These frameworks were considered a product of intervention science and therefore not applicable for practitioners of health promotion. In contrast, MAP-IT was considered timesaving, easy to use and helpful to practitioners when selecting techniques to address relevant mechanisms, thereby reducing the need for in-depth knowledge of all available techniques.
The MAP-IT matrix recognises that personal and environmental mechanisms must be addressed to achieve behaviour modification. The matrix is therefore divided into two main categories; 1) personal
45 which refers to social-cognitive mechanisms and 2) environmental which includes the physical, political/economical and socio-cultural environments (Hansen et al., 2017]. The first column of the matrix outlines the mechanisms present in the personal and environmental categories as identified by theory and evidence-based mechanisms, using a consensus approach. Next, theories underlying the mechanisms and objectives regarding the mechanisms are identified, respectively. The rest of the matrix consists of nine columns, referred to as components, which include the intervention functions as outlined by Michie et al. (2011b) in the BCW. The rows of the matrix link a mechanism to an objective which is underpinned by a theory. Each mechanism is then linked to a BCT as categorised in one or more intervention component.
Strengths of MAP-IT
MAP-IT allows researchers to create a matrix of personal and environmental mechanisms with the aim of promoting positive behaviour. It uses relevant theories and functions of an intervention which are used to address each mechanism. Researchers can then use the matrix to develop an intervention which is underpinned by theory and evidence [Hansen et al., 2017]. The matrix focuses on one aspect of behaviour change when developing the intervention and therefore does not require researchers to have extensive knowledge regarding psychological theory. This knowledge is often lacking in the research team, resulting in the incorrect use of existing frameworks which require understanding thereof [INDEX, 2019]. The matrix also allows scientific research to be linked to practical real-life applications [Hansen et al., 2017]. The MAP-IT approach is considered feasible and low-cost approach to intervention development [Hansen et al., 2017]. Another strength is that it is a synthesis of three well-known and well-respected concepts, namely the BCW, IM and ANGELO [Hansen et al., 2017].
Limitations of MAP-IT
A significant limitation of the MAP-IT approach is that while IM is implemented in the development of an intervention, it only facilitates one aspect of IM, and therefore does not consider a complete approach to intervention development [INDEX, 2019]. The context within which the intervention is also to be implemented is not considered. This is important as an intervention which is systematic, well- designed and embedded in theory will not be effective if implemented in an inappropriate setting.
It is therefore considered to be insufficient in its approach to intervention development [Hansen et al., 2017]. The designers also only offer a single matrix and additional matrices will have to be produced for other conditions or risk factors [Hansen et al., 2017].
46 D. Normalisation process theory (NPT)
The NPT was developed in 2010 by Murray et al. because of problems encountered in the design and implementation of complex interventions. The research group recognised that recruitment into trials was problematic. While identification of the reasons for this phenomenon remain difficult to ascertain, it is evident that trials which form part of existing clinical practice have a greater chance at successful recruitment [Murray et al., 2010]. The researchers also recognised three gaps between research and implementation that needed to be addressed, namely 1) the difficulties and barriers of moving from laboratory-based research to clinical medicine; 2) development of new treatments and dissemination of information regarding the treatments to ensure implementation thereof in practice, and 3) ensuring that the results of successful research are translated into wider health-related policy and practice [Murray et al., 2010].
NPT explains how interventions function from the early implementation thereof and beyond by identifying factors which promote and inhibit the incorporation of interventions into everyday practice.
It has four main components which are considered to be non-linear and in a dynamic relationship with each other. It is an action theory where each construct is used to explain the work that people do rather than their attitudes and beliefs [Murray et al., 2010]. The four constructs are outlined in Table 2.5.
Table 2.5: Description of the Normalisation Process Theory and its four constructs
Constructs Description
Construct 1: Coherence (meaning and sense-making of participants)
Differentiation: Understanding how a set of practices and their objects differ from each other.
Communal specification: Participants work together to understand the aims, objectives and the expected benefits of a set of practices.
Individual specification: Individual participants gain understanding in their specific tasks and responsibilities related to a set of practices.
Internalisation: Participants understand the value, benefits and importance of a set of practices.
Construct 2: Cognitive participation (commitment and engagement by participants)
Initiation: Key participants need to work to drive a new or modified practice forward.
Enrolment: Participants are required to organise themselves to collectively contribute to the work involved in new practices.
Legitimation: Participants need to believe that it is right for them to be involved and that they can make a valid contribution.
47
Activation: Collectively participants need to define the actions and procedures required to maintain a practice and for them to stay involved in it.
Construct 3: Collective action (work done by the participants that makes the intervention functional)
Interactional workability: Interactional work that the participants do with each other and with other components of a set of practices.
Relational integration: Participants build accountability and maintain confidence in a set of practices, as well as in each other, as they use it.
Skill set workability: Allocation of work or tasks involved in a set of practices as they function in a real-life setting.
Contextual integration: Allocation of resources to manage a set of practices.
Construct 4: Reflexive monitoring (formal and informal appraisal of cost and benefits of intervention by participants)
Systematisation: Participants may need to determine how effective a set of practices are by performing a formal (such as a randomised clinical control trial) or an informal (such as a collection of anecdotal examples) systematisation.
Communal appraisal: Participants work together in a formal or informal group to evaluate the value of a set of practices.
Individual appraisal: Participants work as individuals to determine the effect of a set of practices on themselves and the context in which it is set.
Reconfiguration: After the appraisal work of groups or individuals it may be necessary to refine or modify practices.
Strengths of NPT
The NPT components can help to describe the context within which the proposed intervention will be implemented and it uses qualitative and quantitative information to define the intervention [Murray et al., 2010]. The NPT recognises that healthcare involves a series of interactions between multiple individuals and it can help to identify the links between various participants (e.g. professionals, patients or managers) which may be affected by an intervention and can subsequently provide solutions to support these interactions [INDEX, 2019]. The NPT therefore does not only consider the needs of a single group or organisational level, but also takes into consideration the wider system issues [Murray et al., 2010], which is an often a neglected aspect of intervention development [INDEX, 2019].
48 Limitations of NPT
The primary focus of the NPT appears to be to ensure that an intervention can be successfully implemented in practice. It however does not provide detail and guidance regarding how to develop interventions per se [INDEX, 2019].
E. Theoretical domains framework (TDF)
In 2012, French et al. developed the TDF using a consensus process and validation. Psychological and organisational theory relevant to the clinical behaviour change of health practitioners was identified. The result was twelve domains which reflect the main contributing factors that impact the clinical behaviour and behaviour change of practitioners. These domains include knowledge; skills;
social/professional and identity; beliefs regarding capability; beliefs regarding consequences;
motivation and goals; memory, attention and decision processes; environmental context and resources; social influences; emotion; behavioural regulation; and nature of the behaviours [French et al., 2012].
The developers believe that the TDF, when compared to previously accepted models of intervention development, offers an extensive framework that identifies potential barriers to change more adequately and therefore results in a greater range of potential intervention components [French et al., 2012]. The TDF is underpinned by the UK MRC framework as well as two previously published intervention development methods which are theory-based.
The TDF comprises of a four-step approach:
1) Identifying the problem: This step identifies who is required to do what action differently.
2) Assessing the problem: This step utilises a theoretical framework to identify which barriers and enablers need to be targeted.
3) Forming possible solutions: This step requires the identification of the intervention components (BCTs and mode of delivery) that can enhance the enablers and overcome the modifiable barriers.
4) Evaluating the selected intervention: The final step seeks to identify how the behaviour change can be measured and understood.
Strengths of the TDF
The TDF is a four-step method to develop interventions; it is a systematic and theory-based. It is a streamlined approach that identifies the theoretical domains relevant to the problem and then identifies the appropriate BCTs. The TDF is considered a comprehensive guide to facilitate theory- informed intervention development. It is not intended to be prescriptive and allows for adjustment and
49 refinement in accordance with the context and setting of the planned intervention [French et al., 2012].
Researchers have used the TDF as part of their intervention development process to identify theories of behaviour change [French et al., 2012].
Limitations of the TDF
The developers of the TDF acknowledge that there is subjectivity in the TDF process which combines research evidence, matrix mapping and feasibility information. They also acknowledge that using the TDF to develop interventions requires considerable time and resources [French et al., 2012]. The amount of detail provided regarding how to undertake each action is limited [INDEX, 2019].
F. Behaviour Change Wheel (BCW)
In 2011, Michie et al. identified 19 different frameworks of behaviour change in a systematic review and evaluated each of these against a criterion of usefulness, namely; comprehensiveness (the ability to be applied to any intervention); coherence (all categories should be logical, appropriate and consistent in terms of specificity); and categories had to be linked to a model of behaviour (e.g. the Theory of Planned behaviour or the Health Belief Model). The systematic review showed that none of the existing frameworks were comprehensive, three were coherent and only seven could be clearly linked to a model of behaviour. As all the identified frameworks exhibited limitations, Michie et al.
(2011b) developed a new framework namely the BCW which met the criteria of usefulness. Synthesis of the 19 frameworks formed the basis for the development of the BCW which addresses the identified limitations, integrates the common features of the frameworks and is linked to a model of behaviour.
The BCW consists of three layers. The focal point of the wheel comprises of the Capability, Opportunity, Motivation – Behaviour model (COM-B model) which incorporates existing behaviour change models as identified by Michie et al. (2011b) and is used to identify target behaviours for an intervention. The second layer of the BCW constitutes nine intervention functions which are described as activities that can be used to change target behaviours identified by application of the COM-B model. The outer rim of the BCW depicts a layer of seven policy categories that comprise actions responsible authorities may use to enable or support interventions.