What do we want to change?

Audits of practice can often easily identify areas that need improving but need to be followed by work to unpick what factor(s)—individual, institutional and/or external—underpin why a best-practice guideline is not being followed. Is there a lack of awareness or understanding about the evidence or how to apply it? Are time pressures or other priorities a barrier? Where there is a collective responsibility, who is making sure it is actually delivered? Are there local formulary or prescribing policy barriers? Does the guideline fit with local service structures and resources, and if not, what needs to change?

Whom do we need to target?

Implementing a guideline and effecting behaviour change is not something that can be done in isolation. Patients and family members may need to be on board as facilitators; many guidelines depend on the support of interprofessional healthcare teams and sometimes policy-makers. Even at a local level, key individuals in the pathway, including consultants, specialist trainees, allied health professionals and managers, will each have a different knowledge base and need a tailored approach.

How should we go about it?

Implementing new practices requires a change in behaviour, which needs some understanding about what is influencing current behaviour and what might stand in the way of the desired change. There are at least 83 theories of behaviour and behaviour change,3 and a recent Cochrane review identified 15 proposed models to change healthcare professional behaviour alone.4 Common to all is that the starting point should be an assessment of the likely barriers (and facilitators).1 Box 2 summarises common barriers to clinician adherence to guidelines.

There is also general agreement4 that beyond these additional steps we should:

• Select key component(s) for intervention, informed by the identified barriers.

• Use behavioural/educational theory and available evidence to influence the choice of intervention.

• Ensure user engagement throughout, including the feasibility and acceptability of the intervention.

Essentially, “for a behaviour change method to be effective, it must: (1) target a determinant that predicts behaviour; (2) be able to change that determinant; (3) be translated into a practical application in a way that preserves the parameters for effectiveness and fits with the target population, culture, and context”.5

Planning for change requires careful analysis and thought. There is no ‘one size fits all’, but there are several models can help to structure the approach (table 1).

What potential tools are available?

The choice of intervention(s) depends on the goal of intervention, who is the audience and what are the barriers. Sharing new knowledge, learning new skills, raising awareness, addressing habits or altering perceptions of risk may each need different strategies.5 In broad terms, educational/informative approaches can improve adherence to practice, but there is less evidence that these alone will improve patient outcomes. Action and monitoring such as audit/feedback and checklist/reminders can improve processes and practice, with likely but uncertain effectiveness on outcome; good use of media can be important where a persuasive framework is key.6

Table 2 summarises the several systematic reviews on specific methods in this field. The evidence is highly varied, often with methodological problems (eg, using different comparators, lacking definition/standardisation for the interventions, varying in how best to measure the outcome) and often of low quality. There is also likely publication bias that may overemphasise the benefits. Some interventions are rarely used alone but instead as part of a multifaceted approach. At least some of the variability (table 2) likely reflects confounding within and across reviews, so a similarity of observed effects is perhaps not surprising. Most of the studies were powered to detect a 10%–20% difference, though this does not preclude a greater effect in well-designed and tailored interventions. There is little comparative literature—most are studies of one intervention versus none (placebo)7—and it is often difficult to establish a control group.

Educational meetings (interactive, didactic and mixed educational meetings) are commonly used to modify health professional behaviour and are certainly effective although this is likely to be limited to simple behavioural changes.7 The role of local opinion leaders, a role “earned and maintained by technical competence, social accessibility and conformity to the systems norms”,8 is difficult to separate from the nature of the intervention, but likely of some benefit, and certainly widely used by industry whenever new trial or licensing data become available. However, such meetings can be labour intensive, with or without opinion leaders: the reliability and validity of those involved is crucial, and their cost and cost-effectiveness are largely unknown. E-learning is now commonplace, and employers frequently use it for mandatory training. It is at least as effective as other learning methods for improving knowledge, is cost-effective and promotes self-efficacy.9 10 However, most of the literature is in undergraduate populations, with very few evaluating clinician behaviour and none looking at patient outcomes. Whatever the format, interventions with built-in self-assessment and simulation are likely better than those without, and well-designed live and e-simulations can also enhance skills.

Audit and feedback, now embedded in many clinical governance structures, is effective, though more so in some situations than others. Audits undertaken as part of a structured national programme can have a huge impact, especially if accompanied by sufficient funding and organisational change. Thus, the Sentinel Stroke National Audit Programme (https://www.strokeaudit.org/) was a key driver towards the now widespread implementation of hyperacute stroke units and thrombolysis. Similarly, reminders and checklists that prompt healthcare professionals to perform an action can clearly modify simple behaviours. This has been shown in preventative medicine, disease management and prescribing, and in improving patient outcomes in surgical fields.11 The proliferation of guidelines means the checklist can become unfeasibly long: with individual patients seeing multiple members of a care team over time, particularly in chronic disease management, it can become increasingly difficult to track what has and has not been addressed according to best practice. The move to electronic records brings the potential for automatic individualised computerised prompts,12 13 although sometimes the change to electronic records can be less effective than expected.12 Computerised decision support tools can improve practitioner performance, though the impact on patient outcomes is less clear.13 For example, following a root cause analysis into treatment delays for paediatric status epilepticus—which identified substantial contributors were delayed decision-making and uncertainties about timing of next steps—a computerised standard treatment protocol that prompted timed interventions very significantly (>50%) improved the time to the patient receiving both first-line and second-line treatments.14 Similar prompting devices also very successfully supported the delivery of recent clinical trials in status epilepticus.15 16

The mass media is frequently targeted by those wanting to influence patients and professional behaviour. A 2002 systematic review17 showed that using the mass media can consistently change the behaviour of both patients and health professionals. The mass media can be a force for good and harm (giving false expectations, dashing hope, provoking alarm; all of which can be divisive for therapeutic relationships), although it is difficult to measure effect size. Although there has been no more recent review, the more widespread and varied social media influences mean that the media’s influence is probably now even greater.

Considering guideline adherence at the outset

If you are a guideline producer or contributor, consider early on the potential needs and values of your stakeholders, as well as potential barriers to implementation including costs. Factors that improve implementation include creating alternate versions for different purposes, providing summaries of recommendations with evidence, including information within the guidelines to aid intervention design and suggested intervention strategies, and descriptions of ways to monitor guideline use.18 Including such tools does improve adherence,19 although its cost-effectiveness is largely untested.

So what should we do?

If you are now feeling as daunted by the prospect of a service improvement project as we became when reviewing the literature for this review, fret not. The first and possibly most important message is to recognise that while some aspects of a service improvement or guideline implantation project can be done by one or two people, really effecting change needs a more systemic approach. Work as a team, and be sure to involve relevant local clinicians, managers, other health professionals and, where appropriate, patients from the outset. Implementation is more successful when clinicians are themselves engaged in tailoring, reflecting on and evaluating the project, as was seen with the WHO surgical checklists.11 Where time and resources are scarce, consider prioritising one key service improvement project at a time, with subpackages led by different team members with a long-term iterative strategy, rather than multiple individual parallel projects on different topics. Guidelines are not static and evolve over time; they accelerate the process of reducing variation in care, but need to respond to change in knowledge, innovation and accepted practice. Box 3 gives some examples of implementation in practice, and (figure 2) also illustrates some challenges and solutions. Many of the change models are simple, but that does not mean easy—it is easy to underestimate the methodological expertise, time and sustained effort required particularly for planning and reflection.20

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

A: illustrative flow chart. B: bone health example: one consultant led on the audits, creation of local guidance, leaflets and texts for general practitioners (GPs) largely independently between 2008 and 2013. As shown (C: adherence), group improvements were disappointing, with considerable individual variation (range 13%–47% adherence by responsible consultant) despite consensus in the planned approach, and availability of the leaflets/texts across the group. OPD, out patient department; PWE, people with epilepsy.

For employers and organisations, ensure your workforce is educated and informed about best practice guidelines and also about implementing change and design theory. There is much less evidence about how best to influence change in organisations, but an increasing recognition that leaving this to market forces, individual clinicians or local champions is insufficient, and a more strategic considered approach is needed. The recent UK NHS ‘Getting it Right First Time’ (http://gettingitrightfirsttime.co.uk/) drive to improve patient outcomes and expanding interest in design thinking21 are evidence of this.

Identified barriers included lack of understanding of the reasoning, knowledge of which team member should be responsible and differences of opinion about the wording of leaflets. Subsequently, we delivered an educational intervention (lecture, sharing published articles); all members of the group and the local bone health (rheumatology leads) contributed to amend guidance and leaflets; the neurology care group lead formally mandated the revised document as policy. Also under consideration are a computerised bone health prompt incorporated within hospital electronic records and/or checklists on clinic letter templates. We shall re-audit following this.