Osteoarthritis is one of the leading causes of pain and disability worldwide [1,2]. Knee osteoarthritis accounts for nearly four-fifths of the burden of osteoarthritis internationally [3]. In the United Kingdom, approximately 10% of adults have the condition [4]. Osteoarthritis is a long-term condition that does not necessarily deteriorate with time and age [5]. For most patients, pharmacological and nonpharmacological interventions are effective for managing symptoms and improving function [6]. Clinical guidelines recommend that all patients with osteoarthritis should be provided with information, education, exercise, physiotherapy, weight loss advise (as appropriate), and medication. Where these more conservative treatments are unsuccessful in alleviating symptoms, knee replacement surgery is recommended [5].

Total knee replacement (TKR) surgery can be an effective procedure for severe knee osteoarthritis [7,8]. However, it includes risks, with approximately 20% of the patients experiencing ongoing pain or dissatisfaction after knee replacement [9]. There are concerns that TKR is both over- and underused [10], evidenced by increasing provision of TKR surgery in the United Kingdom [11-14], inappropriate referrals [15], and inequity in access to surgery across different geographical regions and social groups [16,17]. The referral rates for TKR are also higher than the uptake of the most strongly recommended nonsurgical treatments (eg, weight loss and exercise) [18,19]. Both patient and clinician factors have been found to influence the over- and underuse of TKR. Patients with knee osteoarthritis have described a limited provision of information about recommended nonsurgical treatment options [20,21] and having a high expectation that they will receive surgery for their condition [22]. Clinicians report that they have inadequate educational materials on alternatives to surgery to share with patients [23] and do not have the time to provide up-to-date information on alternatives to surgery or to address patient values and preferences for treatment [19].

Shared decision-making (SDM) has been proposed in the National Health Service (NHS) Long Term Plan [24] as a way to support patients to make more informed choices about their care while simultaneously reducing inequities in access to treatment [25]. The patient is supported by the clinician to weigh up the risks and benefits of each treatment based on evidence-based information in the context of their preferences and values [26-28]. SDM is well suited to treatment decisions about knee osteoarthritis because decisions are sensitive to preference, and there are several options, each with its own risks and benefits [16]. There is evidence that SDM and the active participation of patients in treatment decision-making can improve clinical and psychosocial outcomes [29-32]. The use of decision-making tools has been found to reduce the uptake of surgery and increase the uptake of more conservative treatment options for several conditions [33-35], including knee osteoarthritis [36-38].

In 2020, the local clinical commissioning group (CCG) commissioned an SDM tool for knee osteoarthritis from another health service in the United Kingdom. Subsequently, a team of musculoskeletal commissioners, academics, and clinicians involved in treating patients with knee osteoarthritis made adaptations to the tool so that it would serve the needs of patients in the local CCG area. The tool aims to prepare patients and clinicians for SDM by supporting them to make informed choices about treatment options and improve patient-centered care planning. The tool is intended to be implemented across all points on the musculoskeletal pathway from primary to secondary care and from early- to late-stage osteoarthritis. It was anticipated that clinicians (general practitioners [GPs], physiotherapists, and surgeons) working across the pathway would use the tool with all patients, regardless of the levels of severity of knee osteoarthritis. The adapted tool was launched in the local CCG area on March 8, 2022. We evaluated the experience of the team adapting the SDM tool and the implementation potential of the tool in the local area. We took a flexible solution-focused approach to this project, shifting from a qualitative study to participatory methodology to adapt to the changing landscape of the NHS while still providing valuable findings to inform the future development of the tool.

The SDM tool was available in web-based and paper versions. Both versions of the tool consisted of a page of instructions for the patient describing how to use the tool, a Helping you decide page that gave patients suggestions about how to prepare for future appointments, a visual diagram, and written information on treatment options. It was anticipated that the web-based version of the tool would predominantly be used by clinicians and patients, with the paper version being an option to be printed out for patients who struggle with digital literacy and digital access. When the user accesses the web-based tool for the first time, they are offered a guided tour. The visual diagram consists of a table that was designed to help the patient determine their osteoarthritis stage, from early to late. The treatment options are described under the headings Self-help, Non-surgical options, Surgery, and Treatments not provided by the NHS. The position of the treatments in the table shows which treatments are appropriate for each stage of osteoarthritis. The patient or clinician can then select the treatments they are interested in and click through for further details. Details for each treatment include the following subsections: What is it?, Benefits, Risks, What happens if no treatment is done?, and Information about support available.

When the tool was launched, clinicians were sent guidelines on how to use the SDM tool and a copy of the paper version of the tool. The guidelines included a link to the web-based version of the SDM tool; suggestions for methods to disseminate it to patients through sharing the link in an Accurx text [39] or in appointment letters or emails; the suggestion that when prior dissemination is not possible (eg, the reason for the appointment is not known ahead of time), the SDM tool could be introduced and shown in the consultation and discussed at a follow-up appointment after the patient has had a chance to look at their options in detail; instructions on how the patient can access and use the tool; and information about how to activate and use an SDM tool EMIS template.

From December 2020 to March 2021, the local CCG conducted 2 one-hour and 6 three-hour SDM training sessions, at the end of which a questionnaire was sent to all attendees asking whether they were willing to pilot the SDM tool. Attendees included GPs, physiotherapists, orthopedic surgeons, and occupational therapists.

The SDM tool was planned to be launched officially in September 2021. However, this was delayed until March 2022 owing to the negotiation of formal collaboration agreements among the stakeholder organizations.

In January 2022, the branding of the adapted tool was agreed, and collaboration agreements were finalized, allowing it to be officially launched in March 2022.

In describing our approach, we have adhered to the COREQ (Consolidated Criteria for Reporting Qualitative Research) guidelines on the reporting of qualitative research [40]. Originally, we had intended to interview up to 20 patients, 20 clinicians, and 5 commissioners about their experience of developing (commissioners) or using (patients and clinicians) the tool. The interviews with the commissioners were planned as patient and public involvement (PPI), whereas the interviews with the patients and clinicians were planned as qualitative research. Because of staff pressures imposed by the COVID-19 pandemic, the recruitment of clinicians and patients was substantially hampered. After discussion with key stakeholders and the study sponsor (University of Bristol), the study design was amended to become a partnership approach using mixed methods [41]. This included (1) a web-based survey to elicit feedback on the SDM tool from clinicians with free-text and multiple-choice responses and (2) a further 3 interviews for the PPI component to explore the views and experiences of the team involved in developing and implementing the tool with the aim of understanding its acceptability and views on its implementation potential. The University of Bristol PPI team provided guidance that anonymized quotes could be used in the dissemination of the work, where consent had been obtained from the people involved in the survey and interviews.

Ethics approval was granted for the original research design by the West of Scotland REC 5 research ethics committee on the 22nd of December 22, 2021 (21/WS/0163). However, ethical approval was not sought for the final research design which used a partnership approach, as it was considered to be PPI.

The methodological orientation used for the interviews was a deductive approach to content analysis [42]. The theoretical underpinning for this qualitative study was the Theoretical Domains Framework (TDF) [43]. The TDF provides a theoretical basis for implementation studies and supports the identification and interpretation of barriers and facilitators to the implementation of interventions in practice [43]. The TDF was developed by a team of behavioral scientists and implementation researchers through a consensus process [44], with 12 domains incorporating 128 theoretical constructs from 33 theories that were judged to be most relevant to implementation questions. After a validation exercise, a 14-domain version of the TDF was developed (version 2.0), which covered 84 theoretical constructs. Version 2.0 has been used in this study [45]. The 14 domains cover individual motivation and capability, as well as factors in the physical and social environment [43].

Clinicians were invited to complete the survey (Multimedia Appendix 1) through snowballing information from musculoskeletal clinical team leaders across primary and secondary care services within the CCG, as well as our research partners in the local NHS Trust and health care providers. Participants were selected for one-on-one interviews if they had been involved in commissioning the SDM tool or adapting the tool for the local area. A purposive sample was sought for the interviews in terms of the role in commissioning the tool for musculoskeletal conditions, adapting the tool, type of clinician, and role in implementation.

The survey was developed in the Jisc online survey system [46]. ST, AJM, and NEW were involved in the development of the survey. The survey was deliberately kept very simple because the work was conducted during the COVID-19 pandemic when restrictions were in place. Because of the simplicity of the tool and the necessity for us to be responsive within the short time frame, no formal validation was conducted. The results were stored on secure University of Bristol servers. A link was circulated via email to potential participants and was available from May 17, 2022, to June 30, 2022. The participants were informed that the purpose of the survey was to evaluate the implementation of the SDM tool. They were informed that providing a response amounted to consent to participate in the study and were asked whether they agreed to the use of anonymized responses in subsequent publications.

Potential participants for the semistructured interviews were approached directly by email and offered a telephone or video call at a time that suited them. At the beginning of each interview, participants were given the opportunity to ask questions and asked for their consent to the interview being audio recorded and to their anonymized quotes being published. Two topic guides were developed, one for commissioners involved in adapting and implementing the tool (Multimedia Appendix 2) and another for those who were using it in their clinical practice (Multimedia Appendix 3). The topic guides were developed by the research team and collaborators and were based on elements from the TDF to include factors that might influence the use and implementation of the SDM tool [39]. The interviews were conducted by 1 researcher (ST) and transcribed verbatim. Field notes were taken during and after the interviews.

For the survey findings, categorical variables were summarized as frequencies and percentages. Interviews were audio recorded with participants’ consent, transcribed, anonymized, and checked for accuracy before being uploaded to NVivo qualitative data management software (release 1.6.2; Lumivero) [47]. Data collection and analysis were iterative and immediate to allow emerging issues to be explored in subsequent interviews. Interviewing continued until we had developed a sufficient level of information power [48], which enabled us to address the aims of the study in accordance with its narrow focus on barriers and facilitators to implementation and the inclusion of a highly specific group of participants.

We followed the guidelines provided in the study by Aitken et al [43] to conduct deductive analysis of qualitative data using the TDF. For both the free-text responses from the survey and the qualitative interviews, content analysis was conducted using the TDF to generate a coding framework [49,50], with quotes being mapped directly to the 14 domains and subdomains outlined in the TDF (version 2.0) [45]. Many of the same themes were coded to multiple domains of the TDF. Therefore, to avoid repetition and to improve the interpretation of our results, we present details of the themes and indicate which domains of the TDF the themes were coded to in the text and in Multimedia Appendix 4. We have also included illustrative quotes for the themes in Multimedia Appendix 4.

Transcripts were read by 1 researcher (ST), who considered the responses in relation to the TDF domains. ST subsequently coded the participant responses to ≥1 of the 14 theoretical domains of the TDF, assigning them to constructs within the domains [43]. A sample of the interview transcripts were read independently by AM and discussed at research team meetings with ST and NW to ensure that the data were coded into the appropriate domains of the TDF and that any subjectivity was reduced. An inductive approach was also used to code responses that did not fit within the TDF domains to ensure that important issues were not overlooked or excluded [51]. Participants were provided with a summary of the findings; none of them responded with feedback.

The qualitative data were grouped according to themes, and the TDF domains are highlighted in Multimedia Appendix 4 and throughout the text. The themes were related to barriers and facilitators to the development of the SDM tool, use of the tool for clinicians and patients, and implementation of the tool. The interview durations ranged from 15 to 61 minutes, with a mean of 37 minutes.

Table 2 provides a summary followed by a description of the quantitative survey findings.

This study aimed to explore the experience of a team adapting an SDM tool for knee osteoarthritis from one health service to another and its implementation potential in the local CCG area. Study participants described barriers and facilitators to both the adaptation of the SDM tool and its implementation and use by clinicians and patients. Facilitators included features of the tool that described treatments appropriate to the patient’s stage of disease, proactive use of the tool, sending it to patients before consultations, using existing NHS dissemination routes to increase awareness of the tool among clinicians, and increasing awareness through clinical leaders and champions. Participants also suggested improvements to the tool’s design and dissemination strategies that could increase its implementation potential and uptake in the future. These themes were mapped to 13 of the 14 TDF domains, with the most salient domains being environmental context and resources and social or professional role and identity. Participants also described usability issues with the SDM tool and improvements that could be made, which did not map to the TDF domains.

Our findings concur with those of Légaré et al [52], whose systematic review of barriers and facilitators to SDM showed that SDM in clinical practice was facilitated by the motivation of clinicians and their perception that SDM would improve patient outcomes and health care processes. The systematic review also reported that across organizational and cultural contexts, the most highly cited barrier was time restraints, followed by a perceived lack of applicability of SDM owing to patient characteristics or clinical situation [52]. In our study, participants often referenced time restraints in their practice. Some of the clinicians interviewed also felt that patients’ engagement with the SDM tool would be affected by particular characteristics, for example, those who were not fluent in the English language or who had lower digital and health literacy levels. The authors of the systematic review cited concerns that these perceptions about which patients would respond positively to SDM and associated tools would result in clinicians only using the tool with selected patients whom they felt might benefit, which would create inequalities in access to SDM and SDM tools [52].

Several suggestions have been made in the literature about how to overcome the challenges of implementing SDM and SDM tools in the NHS [26]. These include making patient decision support available before referral [26], adapting tools rather than reinventing them [26], making tools accessible to both patients and clinicians [53], ensuring that they fit into clinical workflows [53], and having buy-in from the health organization [54]. They suggest that tools that provide short summaries for use in clinical encounters alongside longer sources of information for patients to read in their own time may be more readily adopted by clinicians [26,54-57]. These mirrored suggestions from our participants about how to improve the implementation, use, and sustainability of the SDM tool in the long term.

We took a flexible solution-focused approach to this project, responding to challenges we faced in recruiting participants and a rapidly changing landscape in the NHS by shifting the methods from traditional qualitative research to a participatory approach. We were unsuccessful in recruiting clinicians or patients for interviews as originally planned when advertising the study through clinical networks and musculoskeletal clinical groups. This is unfortunate because it limits the breadth of perspectives from different stakeholders. When we investigated possible reasons for the poor response, clinical leads reported that clinicians were under severe time pressures after the COVID-19 pandemic and were unable to spare time to be interviewed or to recruit patients for the study. After discussion with the study sponsor, the study design was amended to include interviews with commissioners and team members involved in the adaptation and implementation of the tool and a short web-based survey that could be completed quickly by clinicians. We consulted with the project sponsors and the PPI team at the University of Bristol to consider ethical issues around the publication of findings from this partnership approach [58] and sought approval to publish anonymized quotes, which allowed us to provide rich descriptive data to illustrate our findings.

The survey was disseminated widely through the CCG area by snowballing information from musculoskeletal clinical team leaders across primary and secondary care services within the CCG, as well as our research partners in the local NHS Trust and health care providers. Of the 23 survey respondents, 22 (97%) were physiotherapists who had engaged and had had positive experiences with the tool. We may therefore have missed capturing the views of those from different clinical backgrounds and those who were less engaged with the tool. We only received surveys that had been completed; therefore, we do not have additional information about those who may have declined to complete it. The low response rate is also likely due to time pressures after the COVID-19 pandemic. We recognize that the timing of implementation was not ideal, given the huge pressures on the health care system at the time.

This study has implications for future commissioning and adaptation of tools for different regions of the NHS. There were elements of good practice highlighted in this study. The team in the local CCG that commissioned the SDM tool followed the recommendation to adapt an existing tool rather than using resources to develop a new one [26]. The CCG team involved a multidisciplinary team to adapt the tool to ensure that the information was evidence based and would support clinicians and patients to make decisions across the care pathway for knee osteoarthritis [59]. The CCG team also sought to evaluate adaptations made to the tool in this project [59]. However, there were limited resources available to the team that adapted the tool, as well as a lack of specialist intervention development expertise. We would make the following recommendations for teams involved in commissioning and adapting tools from one context to another (many of these are also highlighted in the ADAPT guidance on how to adapt interventions to new contexts [59]):

Researchers may also consider the use of other tools, such as the theoretical framework of acceptability, to support the assessment of the acceptability of health interventions earlier on in the co-design process [63].

This study highlights the barriers and facilitators to adaptation of an SDM tool from one health context to another and its implementation. We recommend that teams involved in commissioning and adapting tools from one context to another should ensure that the tool selected for adaptation has a strong evidence base as well as evidence of effectiveness and acceptability to the target users in the original context. Teams commissioning the tool should seek out legal advice at the beginning of the project and draw up agreements among collaborators to ensure that all parties are clear on expectations and aware of any limitations to changes that can be made. Existing guidance for the development and adaptation of interventions should also be used early on in the process (eg, Medical Research Council and ADAPT guidelines as well as normalization process theory). The accessibility and acceptability of the tool can be improved using co-design methods.