Background

Many resources have been directed toward improving the quality and safety of health care, but major problems persist [1]. Implementation interventions are interventions designed to change clinical practice behaviour and improve the uptake of evidence into practice. To date, implementation interventions have had limited and varied effects [2]. This may be due, in part, to a lack of explicit rationale for the intervention choice and the use of inappropriate methods to design the interventions [3–5].

The design of implementation interventions requires a systematic approach with a strong rationale for design and explicit reporting of the intervention development process [6–8]. One option is to use theory to inform the design of implementation interventions [3, 9]. The UK Medical Research Council’s (MRC) guidance for developing complex interventions informed by theory [10–13] is useful as a general approach to designing an implementation intervention, but it does not provide detailed guidance about how to achieve this.

Multiple theories and frameworks of individual and organisational behaviour change exist, and often these theories have conceptually overlapping constructs [14–16]. Only a few of these theories have been tested in robust research in healthcare settings. There is currently no systematic basis for determining which among the various theories available predicts behaviour or behaviour change most precisely [17], or which is best suited to underpin implementation research [16, 18]. Theories that have been used in previous implementation research include PRECEDE (Predisposing, Reinforcing, and Enabling Constructs in Educational Diagnosis and Evaluation), diffusion of innovations, information overload, and social marketing [5].

This article is one in a Series of articles documenting the development and use of the Theoretical Domains Framework (TDF) to advance the science of implementation research. The TDF was developed using an expert consensus process and validation to identify psychological and organisational theory relevant to health practitioner clinical behaviour change [19]. A set of 12 domains covering the main factors influencing practitioner clinical behaviour and behaviour change were identified: knowledge; skills; social/professional role and identity; beliefs about capabilities; beliefs about consequences; motivation and goals; memory, attention and decision processes; environmental context and resources; social influences; emotion; behavioural regulation; and nature of the behaviours. Relative to previously accepted, often implicit, models for developing interventions (for example, need to raise awareness, to provide information, to educate, need for an opinion leader or champion), these 12 domains provide an extensive framework that has greater coverage of potential barriers to change, and thus implies a greater range of potential intervention components.

Although improved health care can be facilitated at different levels of the health system, one important approach is to support individual health professionals to modify their clinical behaviour in response to evidence-based guidance [15]. The focus on this level is because much of health care is delivered in the context of an encounter between a health professional and a patient, making healthcare professional clinical behaviours an important proximal determinant of the quality of care that patients receive.

Development of implementation interventions can draw on theory, evidence, and practical issues in the following ways. Theory can be used to understand the factors that might influence the clinical behaviour change being targeted, to underpin possible techniques that could be used to change clinical behaviour [19], and to clarify how such techniques might work. Evidence can inform which clinical behaviours should be changed, and which potential behaviour change techniques and modes of delivery are likely to be effective. Practical issues then determine which behaviour change techniques are feasible with available resources, and which are likely to be acceptable in the relevant setting and to the targeted health professional group.

This paper outlines a method for developing an implementation intervention, drawing on the guidance from the UK MRC framework [10–12], and building on previously published methods for theory-informed intervention development [20, 21]. We discuss how we used this method for developing an implementation intervention using the example of a recently developed intervention that we tested in the IMPLEMENT cluster randomised trial [22].

Methods

A method for developing implementation interventions to change clinical behaviour

We used a four-step approach, consisting of guiding questions to direct the choice of the most appropriate components of an implementation intervention (Table 1). The four steps represent: identifying the problem (who needs to do what, differently?); assessing the problem (using a theoretical framework, which barriers and enablers need to be addressed?); forming possible solutions (which intervention components could overcome the modifiable barriers and enhance the enablers?); and evaluating the selected intervention (how can behaviour change be measured and understood?).

Table 1 Steps for developing a theory-informed implementation intervention
Full size table

Step 1. Who needs to do what, differently?

We selected the target clinical behaviours to be addressed, based on documented evidence-practice gaps. We specified the target behaviours in detail by asking the following questions: What is the clinical behaviour (or series of linked behaviours) that you will try to change? Who performs the behaviour(s)? And when and where do they perform the behaviour(s)?

Step 2. Using a theoretical framework, which barriers and enablers need to be addressed?

We chose a theoretical framework that we considered was most likely to inform the pathways of behaviour change. We used qualitative methods, underpinned by this theoretical framework, to identify the barriers and enablers to the pathways of change that were likely to influence the target clinical behaviours.

Step 3. Which intervention components (behaviour change techniques and mode(s) of delivery) could overcome the modifiable barriers and enhance the enablers?

Informed by our chosen theoretical framework, and empirical evidence about effectiveness of behaviour change techniques, we identified techniques to overcome the barriers and enhance the enablers. We first established the content of the intervention (what will actually be delivered), then we identified possible modes of delivery (how each chosen technique would be delivered) [23]. We based the final selection of behaviour change techniques and mode of delivery on what we considered was locally relevant, likely to be feasible, and could be implemented as a cohesive intervention.

Step 4. How can behaviour change be measured and understood?

We determined in advance the outcome measures for behaviour change and which mediators of change could be measured to evaluate the proposed pathways of change [24]. We based the selection of outcome measures on the availability of reliable and valid measures that were feasible to use.

Results

The resultant IMPLEMENT intervention was delivered via two facilitated interactive small group workshops that were a combination of didactic lectures and small group discussions and activities. We also produced a DVD to distribute to all general practitioners (GPs) in the intervention group with the primary purpose of providing the material to those who could not attend the workshops. This alternative mode of delivering the same intervention content included film footage from the workshops and electronic resources related to acute low back pain management.

Step 1. Who needs to do what, differently?

The target behaviours for the IMPLEMENT intervention arose from two recommendations from the Australian evidence-based clinical practice guideline for acute low back pain [25]. The first target behaviour was to restrict the ordering of plain film x-rays to situations in which fracture is suspected because plain film x-rays are rarely helpful in the management of acute low back pain and are potentially harmful. The second target behaviour was to advise patients with acute non-specific low back pain to remain active because this reduces pain and disability.

We chose these target behaviours because they had strong supporting evidence, were potentially modifiable at a practitioner level, and were clinical behaviours to be performed by the GP during a clinical interaction early in the course of management of acute low back pain.

Step 2. Using a theoretical framework, which barriers and enablers need to be addressed?

To develop the IMPLEMENT intervention, we used the TDF [19] to identify the barriers and enablers to the target behaviours and to guide the choice of intervention components. Barriers to, and enablers of, the two target behaviours were identified in a qualitative study consisting of focus group interviews with 42 GPs in Victoria, Australia [26]. Each focus group was led by a trained facilitator who investigated the reasons GPs gave for practising, or not, in a manner consistent with the guideline recommendations. Questions were designed to explore the domains from the TDF for each of the two behaviours. Key domains were identified that described the specific barriers and enablers at a theoretical level, which then allowed us to access relevant evidence of likely effective behaviour change techniques.

Step 3. Which intervention components (behaviour change techniques and mode(s) of delivery) could overcome the modifiable barriers and enhance the enablers?

For the IMPLEMENT trial, our selection of behaviour change techniques was informed by a matrix that mapped behaviour change techniques to the theoretical domains, based on expert consensus about effectiveness for behaviour change [27]. We used the experience of the research team including clinicians and clinician educators, together with feedback from clinical colleagues on potential intervention approaches, to determine which behaviour change techniques and modes of delivery to select.

We chose the delivery mode of facilitated workshops because interactive education is familiar, acceptable, and feasible for GPs, and there is evidence that interventions delivered using this mode of delivery may change professional practice [28]. Also, this delivery mode could be linked to the requirements for Continuing Professional Development points for GPs in Australia. Finally, the intervention was assessed by the clinical members of the research team who checked that the proposed content was likely to be regarded by participants as relevant and helpful to their practice.

Table 2 provides details of the intervention development process of the IMPLEMENT intervention. The columns indicate how we linked specified barriers and enablers to theoretical domains and then identified behaviour change techniques and modes of delivery. For example, one identified barrier was lack of skill: some GPs reported that they lacked the communication skills to reassure patients that a plain film x-ray is unnecessary. We mapped this barrier to the domains ‘skills’ and ‘beliefs about capabilities’. It was considered that these domains were best addressed by using the behaviour change technique ‘rehearsal’ [27]. We delivered this in the facilitated workshop in a small group activity, where the GP took a clinical history with a trained simulated patient who repeatedly requested an x-ray, and the GP was asked to explain to the patient that a plain film x-ray was unnecessary, followed by a group discussion of the issues the GPs faced during the activity.

Table 2 Description of the steps used to choose the behaviour change techniques for the IMPLEMENT intervention
Full size table

Detailed documentation of the full content of the IMPLEMENT facilitated workshops is available in the Additional file 1.

Step 4. How can behaviour change be measured and understood?

Table 3 outlines the constructs we planned to measure in the IMPLEMENT trial, describing outcomes measured to assess the causal pathway (mediating mechanisms of behaviour change), the practitioner outcomes and the patient outcomes.

Table 3 Mediators and outcomes measured in the IMPLEMENT trial
Full size table

Discussion

We have illustrated a four-step method for developing an intervention designed to change clinical behaviour based on a theoretical framework. We have demonstrated the use of this method in a case study of designing the IMPLEMENT intervention, an intervention designed to improve the management of low back pain in general practice [22].

Many researchers who assess barriers and enablers for implementation problem assessment do not do this within a theoretical framework and are therefore limited to pragmatic, rather than theoretically informed, solutions [37]. The use of our method when designing implementation interventions allows the use of theory and empirical research, along with the results of mixed methods research, to decide upon intervention components and to build a complex intervention. This will allow for further exploration of the associations between intervention components and intervention effects [38]. We also encourage researchers involved in developing implementation interventions to better document and report their own development process thus generating a body of knowledge about explicit methods for developing these interventions.

Strengths of this method

The main strength of this four-step method is that it can be used as a guide for implementation intervention developers through a systematic method of moving from target behaviours, to theoretical domains, to behaviour changes techniques, and finally a full implementation intervention. By basing implementation interventions on a theoretical approach to behaviour change and linking this to relevant and effective behaviour change techniques, researchers can make explicit, and thus investigate, the hypothesised mechanisms of change. We propose a streamlined approach moving directly from identified theoretical domains relevant to the implementation problem to behaviour change techniques.

Multiple theories and frameworks of individual and organisational behaviour change exist [14–16], and choosing an appropriate theory for designing implementation interventions is challenging [16–18]. Setting predetermined criteria for selection of theories may assist, for example, including consideration of the evidence base for the theory, the relevance to the setting, and perceived usefulness of the theory. Also, using a broadly based theoretical framework for behaviour change, rather than a single theory, may allow a more comprehensive examination of potential barriers and enablers, and possible mechanisms linking them to the target clinical behaviour. The TDF is arguably the most comprehensive framework for designing implementation interventions as it offers broad coverage of potential change pathways; however other theoretical frameworks, or specific theories, could be used.

There are many potential delivery modes in the clinical setting for most behaviour change techniques [27], including educational meetings, educational detailing, reminder systems, and audit and feedback [39]. We suggest the choice of the delivery mode is guided by local context and what is acceptable and feasible in the target group.

Potential limitations of this method

There is subjectivity in this proposed process of designing implementation interventions of combining research evidence, matrix mapping, and feasibility information. Further empirical work is required to test the method in different contexts and examine the links between theoretical assessment and behaviour change techniques. There are alternative ways of operationalising this method. For example, in other studies the TDF has been used to identify key domains followed by predictive theory-based surveys to check whether the domains regarded as ‘key’ actually do predict clinical practice [40]. Another approach has been to use the TDF to identify specific theories of behaviour change to inform the intervention [40–42]. Documentation and careful reporting of the development process of implementation interventions will further this field.

In IMPLEMENT, barriers and enablers were assessed by conducting focus groups with GPs. Hence, the intervention developed was directed at the individual clinician. However, in broader application of the TDF, although data may be gathered at an individual level, theory is not restricted to the individual level and may address organisational determinants of behaviour change, such as the included domains Environmental context and resources, Professional role and identity, and Social influences which address issues beyond the individual clinician.

This method of intervention development requires considerable time and resources. Developing the IMPLEMENT intervention involved time of a PhD student (SDF), his supervisors (SEG and RB), and input from the rest of the research team via teleconferences and email correspondence over 12 months. During this time we were also developing the method itself, adding to the time needed. Full detail of the resources required to develop the IMPLEMENT intervention will be reported in a separate economic evaluation publication.

The resources required could be seen as a necessary component of the development of complex behaviour change implementation interventions [12]. Resources to enable this will depend on funding models allocating sufficient funds for the intervention development component of research studies and to require this as a component of guideline development and implementation initiatives. Currently, many funding bodies expect that, at the time at which funding applications are submitted, the intervention protocol for complex interventions is already well formulated [43]. This contrasts with, for example, pharmacologic interventions, for which it is accepted that a decade or two of basic research is required before an intervention is ready for trialling. The publication of the new guidance for the MRC framework [12], along with applications of the framework which demonstrate the resource intensive nature of this work, will hopefully facilitate increased funding for this important activity, increasing the likelihood that interventions will be fit for evaluation in trials, and will therefore better utilise research investments.

Conclusions

We have illustrated a four-step systematic method for developing an intervention designed to change clinician behaviour based on a theoretical framework. We propose a streamlined approach moving directly from identified theoretical domains relevant to the implementation problem to behaviour change techniques. This method is a conceptual aid, rather than a rigid prescription; it may be iteratively adjusted and refined to suit other contexts and settings. The process outlined here can be used by researchers and quality improvement practitioners to guide a comprehensive intervention development process.

By basing implementation interventions on a theoretical approach to behaviour change and linking this to relevant and effective behaviour change techniques, researchers can make explicit, and thus investigate, the hypothesised mechanisms of change. We have argued that using such a method can facilitate the development of theory-informed implementation interventions. Finally, appropriate reporting of the processes used to develop the interventions, and of the components of the intervention, is necessary.