In the following paragraphs, we present some strategies that might help to harmonise guideline recommendations with SDM. However, developing a guideline in a way that enables and promotes SDM will require strategic planning by a guideline team who is committed to this concept, oversees the whole guideline production process, and creates a guideline format that is most appropriate to enable SDM for the specific condition or topic in question.

Strategic planning and the provision of additional guideline-based knowledge and decision tools also need resources and expertise. Not every guideline group will be able to adopt all the strategies suggested in this section and they will not be suitable for every recommendation in a guideline. Therefore, the group will need to prioritise which strategies to use and which recommendations to choose. Some suggestions on how to identify such recommendations are given at the end of this chapter.

Based on qualitative interviews with guideline and SDM experts and group discussions with international participants, a team of experts developed a framework of suggestions on how guidelines could be adapted to enable SDM. They characterised these strategies as those:

• aimed at HCPs or patients, or
• generic or aimed at a specific recommendation (see table 1; van der Weijden et al. 2013).

Table 1 Strategies for implementing SDM in guidelines (adapted from van der Weijden et al. 2013)

Strategies within the guideline, aimed at the HCP	Strategies within the guideline, aimed at the HCP	Strategies linked to or within the guideline, aimed at the patient	Strategies linked to or within the guideline, aimed at the patient
Generic strategies	Recommendation-specific strategies	Generic strategy	Recommendation-specific strategies
• Separate chapter on SDM • Language that involves patients	• Cluster 1: Structuring options to increase option awareness • Cluster 2: Structuring the deliberation process	• Patient version of a guideline	• Cluster 3: Providing patient support tools linked to or within the guideline

Some of the suggested elements have generally been shown to be effective, such as decision aids (Stacey et al. 2017); others lack direct evidence and are based on reasoning, experience and expert opinion. The list of strategies is not complete and they should be understood as suggestions that have been found to be helpful and feasible, based on experience in various guideline groups.

In the following sections, we discuss generic strategies and then move on to those aiming at a specific recommendation or topic.

Changing the wording in recommendations

A very simple strategy that highlights the importance of patient-HCP interaction is using wording that encourages discussion and engagement in SDM. For example, using ‘offer’ or ‘recommend’ instead of ‘perform’, and ‘discuss with the patient’ instead of ‘do’. Currently, it seems that several guidelines have adopted this wording, as shown in an investigation of 2 national guideline programmes in Germany (Schaefer et al. 2015).

The GRADE wording of ‘we recommend’ and ‘we suggest’ reflects this idea, although some guideline groups were not altogether comfortable with ‘suggest’ always indicating a weak recommendation. This might not offer enough guidance, because it does not allow differentiation between options that are poorly investigated, and options that have high-grade evidence showing benefit outweighing harms, but with a risk profile or treatment burden such that individual choices are likely to be highly preference sensitive.

Presenting options and their benefit–harm profile in the guideline

This strategy aims to present the options discussed in the guideline in a way that enables HCPs to adequately discuss them with patients. The presentation should also include evidence-based options that may be viewed as second best by professionals (for example, because they are deemed to be less effective), but that may be embraced by patients (for example, because of less intense side effects).

It seems a promising strategy because evidence suggests that some physicians have difficulties in understanding relative risks and adequately communicating those (Wegwarth et al. 2012). It includes:

• listing all the options including no intervention in a comparable format, ideally in tables or graphs
• providing the benefit–risk profile with important outcomes, including treatment burden (Dobler et al. 2018) that allows comparisons of options by:
  • providing absolute effect sizes rather than relative reductions
  • using the same framing for all options and outcomes presented
  • avoiding wording that suggests a value judgement (such as ‘dramatic reduction’ or ‘minimal increase’)
  • highlighting uncertainty (in wording and effect size, with confidence intervals).

For evidence on some of these options see the systematic reviews in Lühnen et al. (2017).

However, a study investigating international guidelines on cardiovascular disease and diabetes found that the vast majority did not provide absolute risk reductions or the numbers needed to treat for interventions (Morgott et al. 2019). Therefore, they did not allow HCPs to grasp the relevance and effect size of the options in question and compare them.

Providing a generic chapter on SDM in guidelines or developing a guideline on this topic

The rationale for a generic chapter on SDM in guidelines is that it could potentially raise awareness of SDM among HCPs, address perceived enablers and barriers to SDM expressed by HCPs, and offer solutions (van der Weijden et al. 2013). There are different examples of such chapters in various guidelines, and they differ in length, content and format. To our knowledge, to date, none of these chapters has been evaluated or tested with guideline users. Therefore, their impact on guideline users remains unclear. Potential downsides of this approach are that:

• its impact may be limited if it is not referred to in the diagnostic and treatment recommendations
• it may easily be ignored by guideline users if it stands separately, and
• it may only be read carefully by those who are already aware of the importance of SDM.

We suggest that if this strategy is adopted, the chapter should not be designed as a textbook. Instead, it should offer practical examples on how to integrate SDM and patient centeredness into treatment planning and evaluation and the examples should be referred to in all relevant recommendations. Discussing treatment goals and planning or evaluating treatment is a core principle of health care provision. Therefore, the concept of goal-based SDM may provide a valuable link between guideline structure and SDM (Elwyn et al. 2020). Box 3 provides an example of a comprehensive generic SDM chapter and its content.

Box 3 Case study on a generic chapter on SDM and treatment planning in the German national disease management guideline on the treatment of type-2 diabetes (Bundesarztekammer 2021b)

Background: Among specialty societies, such as the American Diabetes Association (2020), it is largely accepted that optimal treatment of diabetes requires discussion of individualised disease-specific treatment goals (HbA1c, blood pressure, cholesterol), and continuous evaluation and adjustment of treatment strategies and goals. The German national disease management guideline (GNDMG) on type-2 diabetes provides, to our knowledge, the first model on how to integrate treatment planning and SDM in a guideline. Content: The chapter was structured to highlight the close relationship between goal setting, SDM and evaluation. The guideline panel made consensus-based recommendations and offered practical advice on the following topics: • Agreement on and continuous evaluation of treatment goals: = considering and prioritising fundamental, functional and disease-specific goals (Elwyn 2020). • Risk communication on diagnostic and treatment options: = principles of adequate risk communication in the context of the clinical encounter (Elwyn et al. 2006, German Network for Evidence-based Medicine 2015). • SDM: = presentation of the SDM-Model and suggestion of questions and phrases to enable HCPs’ engagement in SDM (Bieber et al. 2016, Elwyn et al. 2017). • Assessment of contextual factors that may influence prognosis, goals, treatment burden and adherence: = before prioritising goals and planning therapy, assess and consider contextual personal and environmental factors (WHO 2001). • Continuous evaluation of goals, treatment burden and adherence: = if goals are not met, before changing treatment strategies: == assess contextual factors as potential reasons and offer solutions, and == evaluate if individual goals are still valid, and if not, agree new goals. The guideline underwent public consultation. Comments on the SDM chapter highlighted its importance and that it was very helpful. However, others indicated that they felt that, although helpful, it was too complex. (Bundesärztekammer 2021b)

Recently, the NICE guideline on shared decision making (NG197; NICE 2021) issued the first ever, to our knowledge, clinical practice guideline on SDM (see box 4). Many of the advantages and limitations discussed earlier may apply equally to this guideline. However, it has already demonstrated its potential to raise awareness and it is unique in addressing system-related factors that may enable or hinder SDM in practice, referring to the concept of organisational health literacy.

Box 4 Case study on development of a specific guideline on SDM (NICE, NG197, 2021)

Background: Every NICE guideline includes a statement emphasising that it is not mandatory to apply the recommendations and that, although professionals and practitioners should take the guideline fully into account, they should also consider individual’s needs, preferences and values, and make decisions in consultation with them and their families and carers or guardian. NICE has also recommended SDM in several general guidelines, such as on patient experience in adult NHS services (CG138, NICE 2012), medicines optimisation (NG5, NICE 2015) and multimorbidity (NG56, NICE 2016). Topic-specific guidance also often explicitly recommends SDM for specific decisions. But because SDM is not yet routinely practised in the National Health Service (NHS), NICE was asked to produce guidance about facilitating SDM and embedding it in everyday practice. Content: The NICE guideline on shared decision making (2021a) addresses the ‘three legged stool’ of the implementation challenge for SDM: engaging and empowering patients and people who use services; engaging and supporting individual HCPs; and engaging senior managers to embed SDM into healthcare organisations and systems. The guideline covers: • Embedding SDM at an organisational level, including: = making a senior leader accountable for embedding SDM within healthcare organisations = identifying senior HCPs and service users as champions for SDM = developing an organisation-wide improvement plan to put SDM into practice = ensuring that training and development for HCPs in SDM includes specific components = promoting SDM to people who use services. • Putting SDM into practice, including: = supporting SDM by offering interventions at different stages, including before, during and after interactions in which a healthcare decision might be made. = giving guidance on what that support should include. • Patient decision aids (PDAs), including: = how HCPs can make best use of PDAs = how organisations can facilitate use of PDAs by HCPs. • Communicating risks and benefits, including = discussing consequences in the context of each person’s life and what matters to them = giving specific recommendations on how to discuss numerical information with service users. The guideline generated a large number (more than 1,100) of comments at public consultation. These were generally supportive. Regarding the development of the guideline, NHS England and NHS Improvement commissioned the development of the NICE Standards framework for shared-decision-making support tools, including PDAs (NICE 2021c). This will help users assess the usefulness and quality of a PDA and help PDA developers conduct a self-assessment of the quality of their tools and processes. To support implementation of the guideline, Keele University and NICE worked in partnership to develop a free online SDM learning package (NICE 2021b).

Systematically identifying and prioritising situations needing SDM support in the guideline

Some clinical situations described in a guideline will be essential for supporting SDM, but others may be of lesser importance. It may be helpful to systematically identify and prioritise such situations. This helps with structuring the guideline process, raising the guideline groups’ awareness of emphasising SDM in the guideline, and assigning resources to the clinical situations that are key for decision support. The aim is to highlight throughout the guideline those recommendations that are most important when providing recommendation-specific decision support tools.

For setting up a structured process, it is important that a guideline group includes HCPs and patients or lay persons. It may be achieved through a criteria-based group rating or other consultation methods (see the chapter on how to conduct public and targeted consultation). For example, the German Association of the Scientific Medical Societies has systematically developed a very elaborate process to identify recommendations suitable for Choosing Wisely, based on rating criteria with a 4-point Likert scale (German Association of the Scientific Medical Societies 2020). This process can easily be adapted to identify SDM-priority recommendations. Nevertheless, simple surveys among the guideline group or other consultation methods can be equally helpful.

Developing guideline-based patient-directed knowledge or decision tools

When a situation with specific need for decision support has been identified, respective tools should be provided. Van der Weijden et al. (2013) suggested a patient version of the guideline, although awareness of other information formats for patients that may accompany a guideline has increased over time. For more information on patient versions of guidelines, see chapter on how to develop information from guidelines for patients and the public.

However, the concept of patient versions of guidelines is not well implemented. For example, a structured analysis of the German Guideline registry indicated that only 35% of all guidelines provided patient versions (Ollenschläger 2018). Furthermore, many providers use patient versions to give specific information in an easy-to-read manner rather than as a tool that the patient and health care provider can use through an SDM process.

A group of researchers recently presented a framework to characterise and categorise the various patient-directed knowledge tools, including those that may be suitable to supporting clinical decisions (Dreesens et al. 2019). For a detailed description of this framework, see the chapter on how to develop information from guidelines for patients and the public.

Furthermore, quality criteria regarding the production of guideline-based patient-directed knowledge tools have been formulated (van der Weijden et al. 2019). They should inform the development process

Among the suggested formats to support decision making, the framework presented by Dreesens et al. (2019) lists decision trees, pre- or post-encounter patient decision aids, patient versions of guidelines, and encounter decision aids (see table 2). But only encounter decision aids are labelled as engaging SDM, based on the existing evidence (Coleywright et al. 2014, Wyatt et al. 2014):

Table 2 Patient-directed knowledge tool types (adapted from Dreesens et al. 2019)

Purpose	Patient information and educational material	Decision tree	Independent/ pre- & post-encounter PDA	Patient version of guideline	Encounter-PDA
To inform or to educate	+	–	+	+	+
To provide recommendation(s)	–	+	–	+	–
To support decision-making	–	+	+	+	+
To engage in SDM	–	–	–	–	+

Encounter decision aids are short tools designed to be used during consultation (‘point of care’ tools). They may vary in format and may be presented, for example, as Option Grid decision aids, drug facts boxes (Schwartz and Woloshin 2013), or interactive online tools, such as tools for GPs.

The International Patient Decision Aid Standard (IPDAS) Collaboration and NICE (2021c) have published advice and quality criteria for SDM support tools, including PDAs.

Decision aids have proven to be effective. A Cochrane review found high to moderate quality evidence that decision aids improved a variety of outcomes, such as knowledge or risk perception and reduced decisional conflicts (Stacey et al. 2017). However, a follow-up study of this review showed that many of the decision aids included in the Cochrane review, that had been rigorously developed, tested and proven to be effective, were not implemented in routine care (Stacey et al. 2019). The most commonly reported barriers were lack of funding, outdated PDAs, and clinicians disagreeing with use of the PDA. Enablers included design for and integration into the care process. The authors suggested that ‘to improve subsequent use, researchers should codesign decision aids with end users to ensure fit with clinical practice.’ (Stacey et al. 2019). This shows how involving the guideline panel may be helpful in the development of decision aids. The panel would be able to provide a broad range of clinical expertise and practical experience from different health care professions and patients. This would ensure that the guideline and the decision support tools are consistent and complement one another, and encourage uptake of both.

There are several ways to provide decision support tools alongside a guideline. Some possible approaches are described here.   

Check out what already is out there

For some situations, high quality decision tools may exist. For example, see the Ottawa Hospital Research Institute’s repository for English language decision aids and the Option Grids decision aids list. If there is consensus, a guideline panel can recommend suitable decision tools and provide a link to them instead of producing new ones.

Production by an external team

A team of information specialists, researchers and patients can develop decision tools for prioritised situations after the guideline has been published, using its systematic searches and evidence tables as basis for the decision tools. Ideally, members of the guideline group should review the draft for consistency with the guideline and the underlying evidence. The tools should be available for patients but also for HCPs, and be linked to the guideline documents (NICE 2018c).

Content management system-based semi-automated production

Guideline development tools and content management systems, such as MAGICapp or GRADE Pro, enable guideline groups to produce semi-automated electronic decision aids for use in the clinical encounter (Agoritsas et al. 2015, Vandvik et al. 2013). These are produced directly from the datasets of the systematic review and critical appraisal for each guideline recommendation or the underlying clinical question and are available directly through the app or other front ends, and so are linked to the guideline itself. Although representing a helpful tool to support the discussion between patient and clinician, a patient decision aid also needs input from patients or patient representatives, because they can shape the value elicitation statements or narratives that are needed to turn it into a decision support tool.

Production during the guideline development process

After having identified relevant situations for SDM, guideline groups can develop decision tools for the guideline. They can be published as print, PDF, and web- or app-based or interactive online tools.

An advantage of these decision tools is that they are directly linked to the guideline and use the same body of evidence. However, they remain separate tools with additional features for interested guideline users, and do not necessarily raise awareness of SDM among guideline users in general. Only if integrated into and referred to by the guideline itself, do these tools gain importance and awareness, as explained in the next section).

Integration of SDM and decision aids in guideline algorithms and recommendations and provision of SDM tools as part of the guideline

Probably the best way to harmonise guidelines with SDM is to integrate all processes and develop a product that reflects the need for clinical guidance as well as the need for sharing decisions. This means integrating SDM in guideline recommendations and algorithms and putting decision tools at the heart of the guideline itself.

Integrating decision tools into the guideline

Decision tools for clinical encounters can be integrated into the guideline by:

• publishing them as integral part of the guideline rather than separately (for example, as an appendix or supplement)
• cross-referencing and referring to the decision tools in the context of the recommendations in question (either in the recommendation itself or the background information)
• a formal consensus process for all decision tools, indicating formal approval of the guideline panel and therefore the same level of credibility and relevance as recommendations or other guideline elements.

Recommending the use of decision aids

Guidelines make recommendations for HCPs. In clinical situations with high relevance, the use of decision aids can be part of a recommendation. For example, based on expert consensus, the GNDMG on CAD recommends:

‘Before receiving cardiac catheterisation, we strongly recommend the use of the respective patient decision aid (see annex). The consultation and use of the decision aid has to be documented.’ (Bundesärztekammer 2019)

An encounter decision aid was developed for this recommendation, presenting the risks and benefits of all treatment options (medical treatment, stenting, coronary artery bypass grafting) in the form of an option grid. The decision, aid as well as the recommendation, underwent formal consensus. This recommendation was reflected and referred to in the treatment algorithm.

Similarly, in its guideline on urinary incontinence and pelvic organ prolapse in women, NICE (2019a, 2019b) recommends:

‘If a woman is thinking about a surgical procedure for stress urinary incontinence, use the NICE patient decision aid on surgery for stress urinary incontinence to promote informed preference and shared decision making.’

The PDA referred to is a longer format PDA because of the complexity of the decision. It was developed by members of the guideline committee (including patient members) and PDA specialists and was formally consulted on.

Other recommendations that include the use of a decision aid might cover issues such as agreeing on individual treatment goals or evaluating treatment strategies.

Algorithms or decision trees

Algorithms provide a concise and dense overview of clinical decisions on the diagnosis or treatment of a condition. They are among the most cited and best implemented elements of clinical practice guidelines (Vader et al. 2020). The GNDMG on type-2 diabetes (Bundesärztekammer 2021a) provides a treatment algorithm that asks for SDM and the evaluation treatment goals before initiating and before modifying treatment (see figure 1). This is expressed through the symbol of speech bubbles indicating: ‘agreement on treatment goals and therapy strategy using shared decision-making’.

Figure 1 Treatment algorithm for type-2 diabetes based on SDM (Bundesärztekammer 2021a)

Shaping guidelines as unique tools to enable SDM – going beyond recommendations

The Canadian PEER initiative proposes an even more radical approach: They advise the use of so called ‘simplified guidelines’ that do not offer treatment recommendations. Instead, they suggest thresholds for discussing different treatment options with patients, therefore being designed to guide the patient-clinician encounter rather than providing guidance for clinicians only (Allen et al. 2017). So far, guidelines for lipid management in primary care, management of opioid use disorder, and prescription of cannabis in primary care have been developed. The idea that guidelines do not provide general guidance but enable individual conversations challenges the concept of guidelines as understood by most guideline developers and users. Giving it serious consideration might pave the way for future patient-clinician decision support tools that could address some shortcomings of actual guideline development. However, these tools may not be appropriate for all clinical situations and conditions. There will be some indications for which more concrete guidance is needed, and other situations in which enabling discussion is more than adequate.

---