How to identify, incorporate and report patient preferences in clinical guidelines: A scoping review

Abstract Background Clinical guidelines optimize care delivery and outcomes. Guidelines support patient engagement and adherence if they reflect patient preferences for treatment options, risks and benefits. Many guidelines do not address patient preferences. Developers require insight on how to develop such guidelines. Objective To conduct a scoping review on how to identify, incorporate and report patient preferences in guidelines. Search We searched MEDLINE, EMBASE, Scopus, CINAHL, OpenGrey and GreyLit from 2010 to November 2019. Eligibility We included English language studies describing patient preferences and guidelines. Data extraction and synthesis We reported approaches for and determinants and impacts of identifying patient preferences using summary statistics and text, and interpreted findings using a conceptual framework of patient engagement in guideline development. Results Sixteen studies were included: 2 consulted patients and providers about patient engagement approaches, and 14 identified patient preferences (42.9%) or methods for doing so (71.4%). Studies employed single (57.1%) or multiple (42.9%) methods for identifying preferences. Eight (57.1%) incorporated preferences in one aspect of guideline development, while 6 (42.9%) incorporated preferences in multiple ways, most commonly to identify questions, benefits or harms, and generate recommendations. Studies did not address patient engagement in many guideline development steps. Included studies were too few to establish the best approaches for identifying or incorporating preferences. Fewer than half of the studies (7, 43.8%) explored barriers. None examined reporting preferences in guidelines. Conclusions Research is needed to establish the single or multiple approaches that result in incorporating and reporting preferences in all guideline development steps.


| BACKG ROU N D
Clinical guidelines are a fundamental approach for translating knowledge to policy and practice because they synthesize the totality of evidence on a given condition, disease, procedure or therapy, and provide recommendations that support decision-making for healthcare planning, delivery, evaluation and improvement. 1 Guidelines have long been recognized as a means of improving health-care professional behaviour and clinical outcomes across all conditions and settings of care. 2,3 Apart from directly informing practice, guideline recommendations can be embedded in clinical decision support applications 4 or inform the development of clinical pathways that facilitate multidisciplinary teamwork 5 or performance measures that underpin evaluation and quality improvement efforts. 6 Considerable research over many decades has generated insight on how to actively implement guidelines by pre-identifying potential multi-level barriers of use 7 and using that information to choose and tailor implementation strategies from among a plethora of educational, social, organizational and system-level options. 8,9 In concert with implementation strategies, another important strategy for supporting guideline adoption is to ensure that guidelines are implementable, referring to the characteristics of guideline that help end-users apply them. 10 For example, guideline developers can employ the Appraisal of Guidelines for Research and Evaluation (AGREE) Consortium, Institute of Medicine, and Grading of Recommendations Assessment, Development and Evaluation processes to ensure that guidelines clearly describe methods by which they were developed, evidence upon which they were based and recommendations for practice, which all contribute to guideline implementability. [11][12][13] Guidance is also available for developing implementation tools that can be included in or with guidelines [14][15][16] and are proven to facilitate guideline uptake by clinicians. 17 Another way to render guidelines implementable is to formulate the recommendations based on patient preferences and offer guidance to clinicians on how to address patient preferences when applying the recommendations. Patient preferences, defined broadly as 'the desire for specific satisfiers of basic needs' and referring to perspectives, values or priorities related to health and health care, are associated with patient satisfaction with health-care experiences. 18 Patient preferences, collected through means such as interviews, focus groups or questionnaires, or by including patients on guideline development panels, can influence guidelines in many ways. For example, patients articulated concerns unique from clinicians on renal disease lifestyle, psychosocial support, quality of life and outcomes, which shaped the development of a guideline on polycystic kidney disease, 19 and patient value judgments about outcomes associated with palliative chemotherapy influenced panel discussions and informed recommendations in six oncology guidelines. 20 These examples illustrate that patient preferences encompass a range of perspectives that include but are not limited to views on treatment options, benefits and risks, and impact not only on health but also on life in general, and may differ across health issues. Importantly, research shows that guidelines that address patient preferences are more likely to be used because the recommendations reflect patient priorities not identified in published evidence upon which guidelines are based, are aligned with patient values that can differ from those of clinicians, and help clinicians engage patients in discussion and shared decision-making, ultimately leading to higher rates of guideline adherence by patients. 21,22 However, research shows that many guidelines do not address patient preferences. For example, a 2007 survey of 31 international guideline developers found that 58% included patients on guideline panels and 45% surveyed patients regarding preferences. 23 An analysis of 137 guidelines published from 2008 to 2013 found that few described patient involvement in guideline development or included preference discussion tools. 24 Few of 101 American developers evaluated in 2016 required patient involvement on guideline panels (8%), asked patients to review draft guidelines (13%) or offered preference discussion tools in their guidelines (20%). 25 One reason may be that evidence on how best to identify, incorporate and report patient preferences in guidelines is sparse. A synthesis of research on patient involvement in guideline development or implementation published before 2010 found that few studies offered substantial information about the processes or resources they employed, and largely identified challenges such as tension between patient and clinician priorities. 26 Similarly, content analysis of methodological handbooks for incorporating patient preferences in guidelines concluded they provided little detail on how to do so. 27 In the last ten years, there has been increasing emphasis and research on how to achieve patient-centred care by engaging patients in their own care and in planning and improvement activities that benefit all patients such as developing guidelines; hence, more insight may now be available on how to generate patient preference-informed guidelines. [28][29][30] The overall aim of this research was to assemble knowledge that could be widely employed by developers to enhance the implementability of their guidelines as a strategy for supporting guideline use. The specific purpose was to synthesize research published in 2010 or later on how developers can identify, incorporate and report patient preferences in guidelines including processes and determinants (facilitators, barriers) and the potential impact on guideline development processes, guidelines and guideline use.

| Approach
We conducted a scoping review using the most recently generated recommended methods comprised of five steps: scoping, searching, screening, data extraction and data analysis, 31,32 and complied with a reporting checklist specific to scoping reviews. 33 We chose a scoping review over other types of syntheses because it is characterized by the inclusion of a range of study designs and processes or outcomes, which facilitates exploration of the literature in a given field, reveals the nature of existing knowledge and identifies issues requiring further primary study. 31,32,34 Similar in rigour to a systematic review, a scoping review does not assess the methodological quality of included studies and does not assume or generate a theoretical stance. 31,32,34 We did not require research ethics board approval as data were publicly available, and we did not register a protocol.

| Scoping
To scope or become familiar with the literature on this topic, we Together, they drafted eligibility criteria based on the PICO (participants, issue, comparisons, outcomes) framework, which were reviewed and refined by MJA and WBB.

| Eligibility
We included studies in which participants were adult patients aged 18+, or family members or care partners with or without guideline experience; health-care professionals of any specialty or setting of care with or without guideline experience; or developers of guidelines including health-care professionals, managers or staff. The issue of interest were studies that described or evaluated the processes and/or impact of identifying, incorporating or reporting patient preferences in clinical guidelines on any procedure (ie preventative, screening and diagnosis) or treatment for any condition. Preferences referred to personal or clinical needs and values for treatment benefits, risks and outcomes. 18 Studies conducted in all countries and published in English language were eligible. Comparisons, referring to data that were collected or units of analysis, included description or evaluation of single or combined processes for identifying, incorporating or reporting patient preferences, or before-after comparisons, or comparison of two or more processes. Study design included qualitative (ie interviews, focus groups, qualitative case studies, content analysis), quantitative (ie questionnaires, time series, before/after studies, prospective or retrospective cohort studies, trials), multiple or mixed-methods, or programme evaluation studies. Outcomes included but were not limited to exploring participant views, awareness or knowledge about patient preferences or whether/how they should be considered and in what guideline development steps; described processes for identifying, incorporating or reporting preferences; identified determinants (facilitators and barriers) of identifying, incorporating or reporting patient preferences; or assessed the impact of patient preferences on guideline development processes, guidelines and guideline-related products, or use of guidelines.
We excluded studies if they examined the effectiveness of clinical interventions (tests, procedures, treatment for diseases) rather than approaches for engaging patients or patient preferences in guideline development, did not collect or consider patient preferences or did not pertain to guidelines. We also excluded editorials, letters, commentaries, protocols and meeting abstracts. We excluded paediatric studies where parents function as surrogate decision-makers, a scenario that differs from direct engagement of patients warranting separate study. Systematic reviews were not eligible, but references were screened to identify eligible primary studies. Guideline development manuals or manuals specific to patient involvement in guideline development were not eligible because we aimed to describe empirical evidence on identifying, incorporating and reporting patient preferences rather than expert opinion or practices and because prior research showed that such manuals offered little guidance. 27

| Searching
We shared eligibility criteria and exemplar studies identified in the scoping step with a medical librarian and jointly developed a comprehensive search strategy (File S1) that complied with the Peer Review of Electronic Search Strategy reporting guidelines. 35 The search strategy

| Screening
To pilot test screening, CK, YK and ARG independently screened titles and abstracts for the first 25 search results against eligibility criteria and discussed discrepancies and how to interpret and apply the eligibility criteria. Thereafter, CK and YK screened all remaining titles and abstracts. Discrepancies were resolved by ARG. CK retrieved full-text items, which were screened concurrent with data extraction.

| Data extraction
A data extraction form was developed by CK and ARG to collect information on study characteristics including author, publication year, country, study objective, research design, processes used to identify, incorporate or report patient preferences, determinants (facilitators and barriers) and findings. To pilot data extraction, CK and ARG independently extracted data from two articles, and compared and discussed findings to refine the data extraction form. CK and ARG undertook two more iterations of independent extraction and discussion of data from four articles. Then, CK extracted data from all articles, and data tables were independently checked by ARG.

| Data analysis
We used summary statistics to report study characteristics (date

| Search results
A total of 1,965 studies were identified by searches, of which 1,888 were unique items, and 1,669 were excluded based on screening of titles and abstracts. Among 278 full-text articles that were screened, 261 were excluded because they were not an eligible publication type (117), did not pertain to developing guidelines (86), did not focus on patient preferences (38), were not English language (12), were not the target population (5) or were a duplicate publication (3). No additional eligible primary studies were identified in systematic review references. A total of 16 studies were eligible for review ( Figure 1). Data extracted from included studies are available in File S2. 38-53

| Identifying patient preferences
Eight (57.1%) studies employed a single method, and 6 (42.9%) studies employed multiple methods. Two studies included patients as F I G U R E 1 PRISMA diagram guideline development panellists. 38,39 Three studies involved patients as panellists and also identified preferences using systematic review, 43 questionnaire 45 or systematic review plus Delphi consensus process. 49 Two studies identified preferences by focus group, 48,52 and 2 studies employed focus groups plus Delphi consensus process 40 or questionnaire. 41 Two studies identified preferences with interviews, 44,50 and 1 study combined interviews with a questionnaire. 51 One study employed a questionnaire only, 42 and 1 study used only a RAND consensus process. 53 Table 1 maps the way that included studies incorporated preferences in guidelines according to the Armstrong et al conceptual framework of engaging patients in guideline development. 37 No studies incorporated preferences in all or most steps. Included studies incorporated preferences in 6 of 10 possible steps (Table 1). Steps not addressed in included studies were as follows: select guideline development panellists, conduct and/or interpret systematic reviews, update the guideline and take part in evaluating the methods and impact of identifying and incorporating patient preferences. Included studies were too few in number to link any particular single or multiple methods of identifying preferences with incorporating preferences in one or more guideline development steps, or whether particular approaches vary by guideline topic. It did not appear that employing multiple approaches for identifying patient preferences resulted in incorporation of preferences in more steps of the guideline development process.

| Incorporating patient preferences
For example, the study that incorporated preferences in the greatest number of guideline development steps (n = 5 steps) employed only a single approach for identifying preferences, a focus group with 15 patients and 8 carers. 52

| Reporting patient preferences
No studies described if or how the guideline they developed or planned to develop did or would report identified preferences, how preferences influenced the guideline development process or recommendations, or how clinicians can elicit or address preferences in discussions or decision-making with patients.

| Determinants and impact of identifying, incorporating or reporting patient preferences
Determinants are summarized in Table 2. Two studies identified facilitators of involving patients in identifying preferences: both suggested that training in research methods and clinical practice guidelines was needed, 40,46 and 1 also recommended a combination of in-person and virtual meetings. 40 Seven studies identified barriers to identifying patient preferences. Of those, 2 studies noted that it was difficult to find relevant studies that described patient preferences, 43,47 and 1 reported that patients found it difficult to use an online questionnaire to rank recommendations. 51 Nearly all other barriers pertained to involving patients as guideline development panellists. Barriers included finding appropriate patients that represented the larger population and not just their own views, 43,46 institutional review boards that questioned the involvement of patients as panellists, 40 scheduling meetings at a time convenient for patients, 52 lack of understanding among patients about what constitutes a preference, 47 lack of understanding of medical jargon 52 and patients becoming easily overruled by professional panellists resulting in token involvement. 49 One study identified a barrier pertaining to incorporating preferences: lack of clarity on how to weight preferences. 47 No studies explored determinants of reporting patient preferences.

| Impact of identifying, incorporating or reporting patient preferences
Although 10 studies aimed to explore methods for identifying preferences, beyond anecdotally stating how preferences did or would influence guideline development steps, few studies explicitly evaluated the processes or impact of identifying or incorporating patient preferences. Three studies found that patients and clinicians nominated 48 or prioritized different topics. 42,44 One study by Armstrong

| D ISCUSS I ON
This study identified few studies published since 2010 on approaches for generating guidelines that reflect patient preferences.
Of the 16 included studies, 10 aimed to evaluate methods but were largely anecdotal; only one study empirically assessed the impact of involving patients as members of guideline development panels

| CON CLUS IONS
Despite the recognized importance of generating guidelines informed by patient preferences, little research over the last decade has generated guidance on how best to identify, incorporate or report patient preferences. This review identified numerous ways that preferences have not, but could have influenced guidelines, and thoroughly summarized facilitators and barriers, knowledge needed by developers to expand and improve their processes. Further research is needed to establish the single or multiple approaches that lead to the incorporation of preferences in the full range of guideline development steps and explicit reporting of those preferences in guidelines.

ACK N OWLED G EM ENTS
We thank Yalinie Kulandaivelu for independent screening and medical librarian Elizabeth Uleryk for assistance with searching.

CO N FLI C T O F I NTE R E S T
The authors have no conflicts of interest to declare.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data are available in article supplementary material.