Development and evaluation of a behaviour change intervention to increase general practitioner REferral of people with hip and knee osteoarthritis to community-based First-linE caRe (REFER): a mixed methods study

Abstract

Background

Globally osteoarthritis is a leading cause of pain and disability. General practitioners (GPs) have a critical role in the management of osteoarthritis in primary care, yet they also face numerous barriers to referral of people with osteoarthritis to osteoarthritis management programs that provide evidence based first line care (exercise, education, and weight management.) Thus, the aim of this project was to co-develop and evaluate the feasibility of a multi-faceted, theory-based behaviour change intervention to increase GP REferral of people with hip and knee osteoarthritis to community-based First-linE caRe (REFER).

Methods

This project involved a mixed-methods modified exploratory sequential design. Registered GPs or GP registrars with a case load including patients with hip or knee osteoarthritis were recruited in Victoria, Australia. Phase 1: REFER was initially designed by mapping GP-specific referral barriers to the behaviour change wheel. Registered GPs or GP registrars engaged in online, one-on-one semi-structured interviews to explore their learning preferences and refine REFER. Interviews were recorded, transcribed verbatim, and managed in NVIVO. Analyses involved an inductive, thematic approach. Phase 2: REFER was evaluated with a sample of GPs using the Reach, Effectiveness, Adoption, Implementation, and Maintenance Qualitative Evaluation for Systematic Translation (RE-AIM QuEST) framework.

Results

Phase 1: 25 GP interviews identified diverse learning preferences and barriers, including time, cost, and lack of enticing opportunities. Learning facilitators included quick and easily accessible options and earning professional development points. Almost all GPs agreed on including an online, interdisciplinary workshop with additional components (electronic medical record template, web-based toolkit, posters and flyers, booster follow-up session). Phase 2: 27 GPs participated in REFER, with 13 engaging in process evaluation interviews. REFER had high acceptability among GPs who participated and was associated with improved knowledge and confidence in OA guidelines and referral options, with a sub-set of GPs self-reporting increased referral behaviours to community-based osteoarthritis care.

Conclusions

Improved GP knowledge and confidence in guidelines and referral options, alongside increased self-reported referral to first-line care indicates REFER has the potential to improve community-based osteoarthritis management. However, prior to scale-up, work is needed to improve reach and engagement with GPs, and to further refine the intervention.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12875-025-02968-x.

Introduction

Hip and knee osteoarthritis (OA) is a leading cause of pain and disability affecting over 500 million people globally [1]. The immense personal and societal burden of OA includes reduced quality of life [2] and work productivity [3]. Health care costs driven by increasing prevalence and use of total joint replacement as an intervention are considered unsustainable [4]. While total joint replacement improves pain in most individuals with OA, guidelines recommend that this surgery should be reserved for end-stage management, and not prior to trialling first-line care including education, exercise-therapy and weight management for those who need it [5, 6]. Education and exercise-therapy are recommended for all people with hip and knee OA, regardless of disease severity [7]. Exercise reduces pain and improves function [8, 9] and is reported to be cost-effective across multiple healthcare settings [10]. However, up to 2 in 5 people who are referred for orthopedic opinion [11] or waitlisted for total joint replacement [12] have not previously trialed first-line care including exercise-therapy. Individuals who fail to engage with exercise miss a key opportunity to potentially delay or avoid the need for costly surgery [13].

General practitioners (GPs) have a critical role in the management of OA in primary care, both in providing direct patient care and coordinating referral to community-based first-line care including osteoarthritis management programs. A large international survey of 1,512 people with osteoarthritis found that 42% are not satisfied with their OA care [14]. In Australia, three-quarters of people visit their GP first when seeking health professional support to help manage OA [15]. Further, most publicly funded OA management programs that include physiotherapist delivered exercise require a GP referral. The most commonly used management approach for OA by GPs in Australia is pain medication (including opioids) prescription [16, 17]. Although exercise-therapy is recommended across all OA guidelines [7], including the Royal Australian College of General Practitioner (RACGP) guidelines [6], Australian GPs are three times more likely to refer to an orthopedic specialist for surgical consult, than to a physiotherapist who can support exercise-therapy [16]. A multitude of barriers have been reported to influence GP referrals to first-line care for OA, including limited knowledge about and/or undervaluing the benefits of exercise therapy, limited awareness and trust in quality of care provided by physiotherapists, inefficient referral processes, and concerns about out-of-pocket costs to their patients [18–21]. Community-based osteoarthritis management programs have been implemented internationally to facilitate first-line care and provide a clear referral option for GPs caring for people with OA. One community-based osteoarthritis management program, Good Life with osteoArthritis in Denmark (GLA:D^®), provides education and exercise therapy, two key components of first line care. It is offered in ten countries globally including Australia [22]. International and Australian registry data indicates that people with OA who participate in GLA:D^® have improved pain levels, quality of life, and function, alongside reduced desire for surgery [18, 23]. Despite wide-spread implementation in Australia [24], lack of GP referral is reported as a key barrier to patient uptake [18–21].

There is a paucity of research on how to change GP referral practices for OA. One Australian study evaluated the effectiveness of implementing a new comprehensive primary care service model for patients with knee OA [25]. However, GP recruitment targets were not met and associated pain and function improvements were not clinically meaningful compared to usual care. A recent scoping review of primary care-based models of care for OA reported that only 11% of studies focused on enhancing patients’ initial primary care consultation (e.g. with a GP) in the local health system, and only 19% of studies evaluated provider-level outcomes (e.g. GP attitudes, competency, and organization of care) [26]. This review concluded that despite emerging efforts to optimize non-surgical primary care for OA management, there is a need for future work that integrates implementation theories and frameworks and meaningfully involving key stakeholders (i.e. GPs) in the intervention design.

This mixed-methods study describes the development and initial evaluation of a multi-faceted, theory-based, behaviour change intervention to increase GP referrals to community-based OA management. Specific aims include to (i) co-develop and refine (using GP consultation) an intervention to increase GP REferral of people with knee OA to a community-based First-linE caRe (REFER) osteoarthritis program (GLA:D^®); and (ii) evaluate the feasibility of REFER with GPs to inform future evaluation, scale up, or adaptation to other contexts.

Methods

This mixed-methods modified exploratory sequential study design was guided by the pragmatism research paradigm, which is based on investigating real-world problems using multiple sources of data and knowledge to answer research questions [27, 28]. It suggests that knowledge is constructed from interactions between people and their environments [27, 28]. Pragmatism proports the use of multi-pronged approaches to address complex problems, such as the integration of collaborative consultation with research stakeholders [28]. To align with the aims, the study was conducted in two phases (i) intervention development and refinement from March 2022 to March 2023 (qualitative), and (ii) evaluation from March 2023 to April 2024 (quantitative and qualitative) and guided by the Reach, Effectiveness, Adoption, Implementation, and Maintenance Qualitative Evaluation for Systematic Translation (RE-AIM QuEST) framework [29].

Mixed-methods integration followed two strategies. Firstly, research phases were connected [30], with findings from semi-structured interviews in phase 1 directly informing the intervention components implemented and evaluated in phase 2. Secondly, quantitative and qualitative data were collected concurrently and triangulated in phase 2, with integration facilitated by a joint display table to illustrate their unique contribution to the comprehensive evaluation [31]. This study reporting is guided by the Good Reporting of a Mixed Methods framework [32]. Ethics approval was received from La Trobe University (HEC21424). All participants provided written and verbal informed consent.

Phase 1: intervention development and refinement

A preliminary GP targeted intervention was drafted by the research team, including input from a consumer partner with knee OA. The intervention was informed by our previous research examining multi-level barriers and enablers for health systems, health professionals, and patients related to the uptake of guideline-based exercise-therapy for OA [18–21, 33, 34]. General Practitioner-specific barriers were organized using the Capability-Opportunity-Motivation Behavioural Model (COM-B), a core component of the Behaviour Change Wheel framework [35]. This framework was used to systematically identify appropriate interventions targeting each GP-specific barrier. During phase 1 of this study, the drafted intervention (Appendix A) was shared with GPs during semi-structured interviews to seek input on refinement. Phase 1 participant recruitment and data collection have been previously described in detail elsewhere [36], thus are summarized below.

Participants

GP registrars and GPs who worked clinically in Victoria, Australia were recruited via three methods, including by advertisements posted in the newsletter of the Victoria primary care practice-based Research and Education Network (VicREN) [37]; direct contact by VicREN; and personal contacts of the research team. Sample size was not specified a-priori, rather interviews were conducted in parallel with early analysis, and recruitment continued until the research team believed that sufficient information power had been achieved [38]. Participant demographics age (range), gender, years of practice, setting (urban or regional), and frequency of seeing patients with hip or knee OA were collected.

Data collection

One-on-onesemi-structured interviews were conducted with GPs online or by phone, led by a woman who was a registered physiotherapist working in an OA hip and knee service with previous qualitative research experience and training. The interviewer had no prior relationship with the GPs. Participants were offered a gift card ($75 AUD) as compensation for their time to participate. Interviews were audio recorded, professionally transcribed verbatim, and managed in NVIVO software (version 1.7.1). Our interview guide was informed by the Theoretical Domains Framework, a comprehensive framework which synthesizes multiple behaviour change theories, consisting of 14 domains covering psychological, social and environmental determinants of behaviour [39]. It was also reviewed by one consumer with lived experience of OA. Overall, the interview guide (Appendix B) had three broad aims, to (i) explore GPs’ views on OA management (previously reported) [36]; (ii) explore GPs’ professional development learning preferences; and (iii) obtain input on the draft intervention. Iterative discussions with the research team determined when sufficient information power [38] to refine the intervention was obtained, after which two additional interviews were conducted to complete data collection.

Data analysis

Analysis of transcripts began alongside data collection. Transcriptions were reviewed and discussed monthly with the research team allowing for evolution of the topic guide and development of initial codes. One researcher (AG) reviewed and coded all transcripts using an inductive thematic analysis approach [40], while researchers CB and AE each coded half of the transcripts. The research team met six times to discuss the data and to solidify the main themes. All researchers were experienced physiotherapists (range 18–39 years’ experience), one (AG) was a PhD candidate who had attended a 2-day qualitative training course and undertaken previous qualitative research, while the remaining were experienced researchers, including qualitative research (range 4–12 years post PhD).

Phase 2: evaluation

To facilitate an in-depth evaluation of REFER, the mixed methods RE-AIM QuEST framework [29] was used. This implementation science framework guides the integration of qualitative and quantitative methods for a robust evaluation across five key dimensions: Reach, effectiveness, adoption, implementation, and maintenance. Quantitative data was obtained from GP pre-and-post workshop surveys, while qualitative data was collected via post-intervention GP interviews.

Our application of RE-AIM QuEST framework was operationalized using an iterative approach. We a-priori defined each dimension of the framework and this was refined during thematic analysis (Appendix C).

Participants

Participant inclusion criteria remained the same as phase 1. GPs from phase 1 were invited to participate in phase 2, with additional GPs recruited via snowball sampling [41] and word of mouth. The a-priori sample size for this feasibility study was 25–30 participants. In addition to following recommendations made by methodological guidelines for feasibility studies [42, 43], this sample size was determined to be sufficiently large to generate meaningful findings to inform a future larger trial, as well as being pragmatic based on the research teams resources and funding.

Data collection and analysis

Quantitative

GPs initially attended a 90-minute online workshop. Before attending, GPs completed a questionnaire that included demographic and practice details (6 questions); knowledge of OA management and guidelines (6 questions); confidence and beliefs about their practice approach and referral behaviours (8 questions); and beliefs about effectiveness of different knee OA treatments (9 questions) (Appendix H). Following the workshop, GPs repeated all questions on OA knowledge, confidence, and effectiveness, as well as answering an additional 5 questions on the value and quality of the workshop. All data was summarised descriptively using frequency (percentage) and mean (standard deviation) or median (range), as appropriate. OA knowledge questions were descriptively summarised using percent change in correct responses from pre-to-post workshop. Responses to pre-post workshop knowledge, confidence and effectiveness questions were dichotomised for statistical analysis: correct versus incorrect; agreement (strongly agree, agree) versus other responses (neither agree/disagree, disagree, strongly disagree); or perceived to be effective (very effective, moderately effective, a little effective) versus other responses (not at all effective, not sure). Data were analyzed using McNemar’s test for paired data with Yates correction to determine significance of changes between pre- and post-workshop.

Qualitative

General Practitioners were asked to participate in a one-on-one booster session to provide an opportunity for individualized support, to meet local GLA:D^®physiotherapists, and to answer any study questions. Additionally, GPs were asked to provide feedback on the behaviour change intervention to the research team as a process evaluation of this complex intervention [44]. Questions were guided by the RE-AIM framework (Appendix D). Sessions were conducted by a member of the research team (AE or AG), held in-person or online (based on GP preference), recorded, transcribed verbatim, and managed in NVivo software (version 1.7.1). Participants were offered a gift card ($75 AUD) for participating. Data analysis began concurrently with data collection. Transcripts were reviewed, checked for accuracy, and corrected as required, alongside the audio recordings. Two researchers (AE and AG) used thematic analysis [40] to inductively code all transcripts. They met three times to discuss, come to agreement on codes and themes, and deductively map the themes to the RE-AIM framework with a third researcher (CB). All researchers were physiotherapists and experienced in qualitative methods.

Results

Phase 1

GPs were recruited from 4.3% of clinics contacted (Appendix E). VicREN indicated that practice managers reported GP workforce shortages, a high demand on daily operations, and a lack of interest as reasons for the low recruitment rate. Twenty-six GPs consented to be interviewed, with one declining audio recording of their interview, resulting in 25 included transcripts. Mean age was 47 years (SD 11), with most (72%) having at least five years of clinical experience. Nearly all GPs worked in metropolitan practices (88%) and saw patients with OA greater than 5 times a week (Table 1). Average interview time was 29 min (range: 15 to 52 min).

Table 1.

Demographics of GP participants in phase 1 and phase 2 *1 GP did not provide data

Characteristic	Phase 1 (n = 25)	Phase 2 (n = 27)
Gender, woman n (%)	11 (44)	14 (52)
Age, years mean (SD)	47 (11)	47 (10)
Practice setting n, %
Metropolitan	22 (88)	25 (93)
Regional	3 (12)	2 (7)
Years of practice n, %*
< 5 years	6 (24)	5 (19)
5 − 10 years	4 (16)	8 (30)
11 − 20 years	6 (24)	7 (26)
> 20 years	8 (32)	7 (26)

Learning preferences were thematically summarised into four main categories (Table 2): (i) diverse professional development preferences; (ii) online searches to answer clinical questions; (iii) learning barriers including time, cost, competing education priorities, and lack of interest in topic; and (iv) learning facilitators including quick and easily accessible options and earning professional development points.

Table 2.

Phase 1 qualitative themes and supportive quotations

General Practitioner learning preferences
Theme	Supportive quotations
GPs had diverse array of preferences for how they wish to obtain professional development. Including, (1) in-person interactive workshops, seminars, or dinner meetings; (2) online live or self-paced webinars; (3) podcasts; (4) research articles.	“I think my preferred learning is still the face-to-face talk, a seminar.” GP 9
	“Look, I do quite like webinars. I think they’re really good.” GP 6
	“Podcasts would probably be good. You can listen your own time. You can listen in the car so just to kill time. And it’s quite flexible, you can just listen to any time.” GP 25
	“Good [systematic] reviews…I think are the most effective ways of getting things that are quickly provided.” GP 22
GPs commonly sought answers to clinical questions through online searching. This included searching within trusted sources such as HealthPathways Melbourne (local health region resource); UpToDate (Wolters Kluwer resource), looking at clinical practice guidelines, or general internet searches.	“I might look up quickly, if I have someone comes in in front of me that I’m not familiar would be HealthPathways.” GP 14
	“I have UpToDate…it’s a massive database and anything I wanna learn on the spot and I just go there.” GP 13
	“I’m a lover of guidelines … something where it’s a management approach ‘cause you have so many conditions that you have to think about, it can be easy to forget the specifics. GP 3
GP learning barriers were time and energy, cost, competing education priorities, and lack of interest in specific topic.	“It’s not just a time, they can be a really good resource. But I must say that there are so many things that I always think, “I really wanna look at that,” but I never get to do it. That can be a problem. GP 16
	“Financial pressures…medicare freezes didn’t help. There’s so much pressure on us.” GP 24
	“The difficulty in general practice is that there are so many areas that we need to become proficient in and maintain proficiency, that there’s a barrage of information coming our way.” GP 22
	“I think with osteoarthritis, a lot of GPs are just – unless they’re really into it, I think that’s gonna be hard. because arthritis isn’t my thing, it doesn’t mean that I don’t see it but I mean all GPs are gonna see arthritis. GP 1
GP learning facilitators were access to quick and easy resources and opportunities to earn continuing professional development points.	“If it’s something that happened to have popped up, and I’ve got a spare moment, I’ve seen one of my tabs that’s open in my open in my search, I go, “Look, there’s that article that was suggested to me by that email,” but I’ll listen to that or read that now.” GP 5
	“Provision of CPD points or some other enticement is helpful”. GP 15
General Practitioner consultation on intervention development
Summary of key findings for draft intervention components	Supportive quotations
Workshop: format GPs had heterogenous preferences for an online or in-person workshop. Online was viewed as convenient, but some felt fatigue from an abundance of virtual learning opportunities. In-person was appealing for the ability to include a practical component. A self-directed format was also suggested by one GP.	“Yeah, online meeting, that’s probably the most convenient.” GP 10
	“If it’s a workshop in-person, it works… this makes us all more practical…rather than just reading and looking or what is that, so you can feel and learn it quick.” GP 8
	“I think it’ll be good, even online would be useful, like something you could do in your own time would be easy.” GP 3
Workshop: Time of day and duration GPs had a range of preferences for workshop timing, including at lunchtime, evenings, or on the weekend. Overall evenings were the preferred time. Some GPs thought 2 h was reasonable time, while others thought it was too long	“Probably after hours. You don’t know what can run over, like emergencies and sometimes you work through over lunch break because things come in. So I think evenings, we don’t have any commitments, I think that might be easier.” GP 25
	I think [lunch time PD sessions] do work because you don’t have to make time out of your week to go…without having to set aside a separate time for it all- to go somewhere for a workshop…it’s a bit more time efficient and – yeah, coming to your [clinic] at lunchtime is quite easy. GP 1
	“That’s pretty good on one topic. Two-hour is reasonable.” GP 13
Workshop -Content GPs generally wanted pragmatic content related to: improving their understanding of the benefits of exercise and how to communicate this to patients with OA; pain management strategies; what do physiotherapists practically do with patients; and specific logistical details about the GLA: D program. They did not want background content that described OA characteristics or prevalence. There were mixed comments about including patient experience content.	“I would love to understand what benefits patients can get through exercises, so I can convince them, because sometimes, they question, “Ah, would it do something or do anything that it would be beneficial?” GP 7
	“Pharmacological management of osteoarthritis would be my interest.” GP 4
	“You refer, but you don’t really know what happens. I actually find it useful to be able to say to patients “Oh, this is what happens when you see the physio. They do XYZ.” GP 3
	“We hear the story every day from the patients, so we don’t want extra stories.” GP 18
Toolkit with focus on patient resources GPs were interested in trustworthy resources such as online toolkit, but there was some concern that it will not be used/will be forgotten during day-to-day practice. Some GPs wanted resources embedded directly within their medical software. Simple resources to share with patients (links, printable) seen as valuable.	“Finding reliable resources and knowing where to look.I can see La Trobe, [is involved in this toolkit], I trust La Trobe, so that’s good.” GP 23
	“I think the only possible downside to [a toolkit is] that they’ll be used while you think of it, but if you haven’t seen or used it for some period of time, it’s easy to be forgotten but potentially quite effective tool…I think that could be quite helpful.” GP 15
	“Yeah, within one page would be the best, the critical important one or you can give a patient a link. If they’re happy, if they want to read more, they’d go for the other link.” GP 20
Patient flyers and posters Most GPs seem to think posters for wall or flyers in waiting room would be useful. Although some GPs thought they were wasteful and/or preferred digital posters or videos.	“People read the things that are on the walls in waiting rooms and spent a lot of time there, so I think that’d be really helpful. I like that sort of thing even in my room.” GP 10
	“If you leave marketing materials in the waiting room for the patients, it is helpful for them.” GP 13
	“Don’t waste your money on flyers and things, they’ll go straight in the bin if they come into our practice.” GP 14
Referral template Most GPs were supportive template and are familiar with embedding these into their EMR. Importantly, they wanted it to be brief.	“That [EMR template] would be good. Most definitely.” GP 19
	“I think so. But I guess I would hope the template is designed in a very user-friendly way.” GP 9
Communication with physiotherapist Almost all GPs stated they appreciated communication back from physiotherapists after a referral. They preferred individualized patient outcomes, rather than generic letters. They also liked physiotherapists to make recommendations for further treatment if needed. Being connected to local physiotherapists was seen to be an appealing aspect of the intervention.	“That’s good because I want to know the feedback. Any referrals or any clinical findings is really interesting actually. It’s really good, so I’m always positive for that.” GP 21
	“I would really like that, yep. I always read letters … also just knowing so that when they come back and see me, that they have done the program. If it’s not working, then yes, definitely time to refer to a surgeon.” GP 23
	“At the conclusion of the [physiotherapy] visits a summary of where they’re at and recommendation of whether they think anything might be more helpful.” GP 16

GP General practitioner, OA osteoarthritis, GLA:D Good Life with osteoArthritis in Denmark, OAHKS Osteoarthritis hip and knee service, EMR electronic medical records, RACGP Guideline Royal Australian College of GP Hip and Knee Osteoarthritis Guideline

Key findings regarding feedback on the draft behaviour change intervention included heterogenous perspectives about its various components (Table 2). Almost all participants agreed on the inclusion of an interdisciplinary workshop focusing on understanding (i) exercise benefits for patients with all severities of OA; (ii) details about physiotherapy management of OA; and (iii) GLA:D^® program details (i.e. program components, specific exercises) and logistics (i.e. how to refer, cost). Most GPs preferred a live workshop, although some favoured the flexibility of a self-directed workshop. Most GPs supported the inclusion of additional intervention components including an embedded electronic medical records (EMR) template; a web-based toolkit with patient resources; GLA:D^® posters and flyers; a physiotherapist communication template; and an individualised GP follow-up session. Further detailed qualitative data for phase 1 is presented in Appendix F. The final REFER intervention is summarised in Table 3.

Table 3.

Final REFER intervention: Capability-Opportunity-Motivation behavioural model (COM-B) barriers to GPs referral of people with knee OA to exercise-therapy mapped to behaviour change wheel (BCW) intervention functions and the REFER intervention component

COM-B	GP Barriers	BCW Intervention Function	Final GP intervention components
Capability	Psychological: GPs do not always have complete knowledge of the current evidence or clinical practice guidelines recommendations for knee OA management. They can undervalue the benefits of exercise-therapy for people with knee OA and some believe that exercise can be harmful for the joint or less effective as other treatments (e.g. medications, surgery). Many GPs do not know where to refer a patient with knee OA to receive a high-quality exercise-program. GPs are not always confident to discuss first line care options with patients or know how to determine if someone with knee OA has already adequately trialled exercise-therapy to warrant further referral (e.g. surgery).	Education, training, enablement, modelling	Online workshop delivered by interdisciplinary research team held online from 7–8:30pm local time. Minimum of 3 email reminders/confirmations sent. GPs who had returned a signed consent form were sent 8 email invitations before being considered non-responders (n = 3). Self-directed workshop option also piloted.
			Workshop content included: RACGP Guidelines, diagnosing OA and communicating to patients, first line care including benefits of exercise, physiotherapy programs including where/how to refer (GLA: D Australia, OAHKS), pain management, surgical management, patient experiences with exercise.
			Individualized booster session (30 min) at GP clinic or online/phone to answer questions and get further feedback on intervention, GPs emailed up to 4 times to engage
Opportunity	Physical: GPs lack access to patient educational resources on knee OA (flyers, posters, or web-based printable infographics). GPs do not have an integrated electronic medical record (EMR) referral template to use for referring patients to physiotherapy.	Environmental restructuring, enablement,	Posters and flyers on GLA: D Program, including the safety and benefits of exercise and patient outcomes
			Web-based GP toolkit with information about OA burden and evidence summaries, how and costs of referring to OAHKS and GLA: D, printable online patient resources, exercise benefits, GLA: D information, and interactive aspects (e.g. video clips, quizzes, infographics)
			EMR embedded referral template to facilitate physiotherapy-led service referrals (OAHKS and GLA: D), printable list of local clinics, online GLA: D interactive map
	Social: GPs describe patients’ expectations include seeking imaging and orthopaedic surgeon referral, and pain medication prescription. GPs are not always well connected with local physiotherapists in community.
Motivation	Reflective motivation: GPs want to see good outcomes in patients following referral to exercise therapy, yet they do not always receive communication from physiotherapists, and they are concerned about the variability in quality of care patients receive. GPs do not believe that exercise is able to delay surgery for their patients with knee OA.	Enablement, persuasion, education	Sample communications template GP would receive from GLA: D or OAHKS physiotherapists

COM-B Capabilities, Opportunities, Motivations – Behavioural Model, GP General practitioner, BCW Behaviour Change Wheel, GLA D Good Life with osteoarthritis in Denmark, OAKHS Osteoarthritis hip and knee service, RACGP Royal Australian College of General Practitioners

Phase 2

Twenty-seven GPs participated in the initial evaluation of REFER, with 19 (70%) having engaged in phase 1. Demographic details are summarized in Table 1. Of these GPs, 22 (81%) attended a live online workshop and 5 (19%) completed an online self-directed workshop. One GP dropped out after attending the workshop and did not provide post-workshop data. Seven GPs requested printed materials related to GLA: D^® and one office manager sought assistance from the research team to install the EMR template.

Quantitative results

Overall knowledge of OA management and clinical guidelines improved following the intervention period, with a greater proportion of GPs answering each question correctly at post-workshop compared to pre-workshop in all but one question (Table 4). Beliefs and confidence related to OA management improved in 5 (63%) questions following the workshop (Fig. 1). Beliefs about the perceived effectiveness of different OA treatments did not change (Fig. 2).

Table 4.

GPs knowledge of OA management and clinical guidelines

Knowledge evaluated	Pre- workshop % correct	Post-workshop % correct	Pre-post workshop % change	McNemars p-value
Physical exercise is not a risk factor for developing osteoarthritis.	82%	96%	14%	2.25
				0.13
Exercise, patient education, and weight loss (where relevant) are the first line treatments for hip/knee osteoarthritis.	78%	92%	14%	1.50
				0.22
A diagnosis of knee or hip osteoarthritis can be made based on an overall assessment of symptoms, clinical findings and risk factors.	100%	92%	−8%	0.50
				0.50
People with osteoarthritis can benefit from therapeutic exercise regardless of pain or radiographic severity.	78%	96%	18%	2.29
				0.13
Oral NSAIDS are conditionally recommended for osteoarthritis pain relief in the RACGP guidelines.	11%	46%	35%	5.82
				0.01
Opioids are strongly recommended against (not effective and risk of harm) in the RACGP guidelines.	52%	77%	25%	5.14
				0.02

NSAIDs non-steroidal anti-inflammatory drugs, RACGP Royal Australian College of General Practitioners

Fig. 1.

GP confidence and beliefs about their OA management approach and referral practices

Fig. 2.

GP beliefs abotu effectiveness of different OA treatments

Qualitative results

Process evaluation interview findings from 13 GPs (54% women), including 3 self-directed workshop participants, are outlined in Table 5. The mean interview length was 28 min (range 23 to 37 min). Additional data is provided in Appendix G.

Table 5.

REFER process evaluation: key themes and supportive quotations

Reach: Understanding GPs who participated in the intervention
THEME	SUPPORTIVE QUOTATIONS
GP motivations for participating in REFER: GPs enrolled in REFER to increase their knowledge about osteoarthritis and to learn more about GLA:D.® Many were already aware of the program. They were also motivated to obtain CPD points, by the financial incentive, and to be involved in research.	“We need to get points. So, all the educational activities are always very welcome… and, you know, the fact that it's the only educational activity that we actually get paid for so like, duh.” GP 26
	“We're aiming for an ideal world where people get referred to an evidence-based exercise program for their knee arthritis and we're currently very far from that, and so I wanted to be involved for that reason, and to increase my knowledge of the various programs…and then just general interest in research.” GP 1
GP Practice details: GP practices were highly variable: Frequency of OA patients seen by GPs’ was highly variable; patients ability to afford private physiotherapy was highly variable; and patient attitudes, beliefs about exercise, and overall health literacy was also variable.	“Most of my patients can actually afford to pay. The part of the thing for me is actually getting them there.” GP 17
	“Oh, well physio is expensive,” they don't wanna spend the money on it especially if it's not covered by a GP management plan.” GP 3
	“So most patients don't want to have surgery, they do want to do the exercise.” GP 11
	“Most of the time when the patients with osteoarthritis come in, they are under the impression that only the tablets or a replacement is the go-to.” GP 2
Effectiveness: Impact of REFER on GPs’ OA knowledge, confidence, and referral behaviour
THEME	SUPPORTIVE QUOTATIONS
Perceived effectiveness of REFER on GPs’ OA knowledge, confidence, beliefs, and referral behaviours: Most GPs reported feeling more confident and knowledgeable in managing knee OA, including when and how to refer to the GLA:D® program. Many stated that knowing more about the structure of the GLA:D® program, helped them to be able to talk to patients about the potential benefits and value of the program compared to general exercise. For some GPs who were already referring to GLA:D®, they stated that GLA:D was now more ‘top of mind’ when seeing OA patients, whereas other GPs self-reported that REFER had ‘changed their practice’ that they were now referring patients to GLA:D®.	“I felt more confident in helping to manage hip and knee osteoarthritis. I felt like I've had something that's evidence-based that I can encourage patients to do, that's also not just medication-based.” GP 3
	“Instead of just telling people, go and see your physiotherapists and just, you know, do knee strengthening exercises. I can now tell them, “Look, would you consider the GLA:D® program where, you know, you've got this prescribed exactly, you know, a program of exercises that have been found to be very beneficial.” GP 26
	“I think [REFER] gave me a lot of really good ideas about how to manage OA in a stepwise manner..Knowing when to refer, like timelines for how long to trial that for, before you refer to a surgeon.” GP 27
	“[REFER] certainly made me think differently…I just assumed that physiotherapy was kind of standard, you know, across the board. But to know that there is actually a specific program…to know that it's specifically tailored for OA, yes, it does definitely change my practice.” GP 28
	“I've looked up GLA:D® physios [on GLA:D website] and have referred specifically for knees to them.” GP 11
Adoption: Willingness to participate in REFER
THEME	SUPPORTIVE QUOTATIONS
GPs’ acceptability of REFER intervention: All GPs (except one) reported REFER to be highly acceptable across all acceptability domains. They spoke positively about REFERs flexibility and usefulness, low burden, aligned with their values, easy to understand, and they felt like they were confident to make the referral to GLA:D® when appropriate. One GP did not find REFER acceptable and perceived it to be ‘out of touch with reality’.	Affective attitude domain: How GP feels about the intervention “It was a good little refresher…It probably just reaffirmed what I already knew…I would encourage colleagues to take the course, particularly those that aren't very confident in their management of osteoarthritis.” GP 10
	Burden domain: The perceived amount of effort from GP that is required to participate in the intervention
	“REFER is not too time consuming, like, you know, I could read that on my own and then the follow up with you wasn't that strange. So I actually didn't think it was that hard to do it, to be honest. “GP 26
	Ethicality domain: The extent to which the intervention has good fit with the GPs value system
	“I think the [REFER] study, it's targeting a really important aspect of healthcare that probably doesn't get as much attention as some other illnesses but I think really important.” GP 3
	Intervention Coherence domain: The extent to which the GP understands the intervention and how it works
	“My management will continue to be just, sort of, diagnoses, talk about the conservative measures and exercise and sending them off to physio.” GP 5
	Opportunity costs domain: The extent to which benefits, profits or values are given up by GPs to engage in the intervention
	“It was useful to sort of go through what you guys are offering [with REFER].” GP 26
	Self-efficacy domain: The GPs confidence that they can perform the behaviour/s required to participate in the intervention
	“I feel more comfortable pitching it to people that they should try GLA:D® program or something similar before they go to surgery.” GP 27
Implementation: Fidelity to use of REFER intervention components
THEME	SUPPORTIVE QUOTATIONS
Perceived facilitators for GPs implementing REFER: included it being flexible, GPs having a special interest in OA and pre-existing relationships with local physiotherapists, and patients having insurance or being able to afford private physiotherapy.	“For other GPs just learning their local pathways and who does [GLA:D®].” GP 10
	“The patients I did refer [to GLA:D®], usually they'll do self-funded because we're in a fairly affluent area. People are pretty happy to just pay or use their insurance.” GP 3
Perceived barriers for GPs implementing REFER: included GPs lack of time, having to manage complex public health funding schemes, lack of communication from physiotherapists, not having a nearby GLA:D location, not seeing many patients with OA, and patient beliefs or multiple health issues.	“If I am referring them and they are sent back because they don't have My Aged Care accessible funds, what should I do? Do I tell them that they will have to pay out of pocket to follow the same plan since it will be helpful for their osteoarthritis of knee and hips or do I just tell them just wait until they get older and access the My Aged Care? I don't know what to do then.” GP 2
	“I did not ever receive any, any, any, any response from all referrals I have made. So for me it is very bad. I'm just a referral body and I have nothing.” GP 18
GPs use of different REFER components Workshop component-live and self-paced: All participants except one GP spoke positively about the workshop, both live online and self-directed delivery modes. GPs valued the ability to ask questions at the live workshop, yet the flexibility of the self-directed workshop was also appealing. Some GPs suggested they would struggle to complete the self-directed option due to competing time demands and the lack of firm deadline.	“The education workshop, I thought was really excellent…I think that that probably was the part that probably has made the most difference.” GP 1
	“I felt like the [workshop] was really useful. I think that was really helpful to learn about the management of hip and knee osteoarthritis and also learn about the GLA:D® program and what it involves and how it benefits patients. I think that's probably been the most useful.” GP 3
	“As a mum with three kids and a scout leader and a GP, your time is ridiculous, so being able to do it on my own time’s really good.” GP 5
	“If you tell me, “Oh, you need to come to this thing on Tuesday at 7 PM,” okay, I'm probably gonna be annoyed that I have to go to this thing on Tuesday at 7 PM but I'll go and it'll be done, whereas if you say, “Oh, I've got the self-directed, and you can do it over the next six weeks, and you have to complete these online modules,” the last time I did one of those, I was so late at submitting it, because I just didn't prioritise it.” GP 1
GLA:D® online map: Nearly all GPs reported using the online GLA:D® map to find program locations.	“I’ll say, “Here's the website. It works.” So I tell them about the evidence of people who are waiting for their hip and knee replacement, I should say, and then they think, “This sounds really good,” and off they go.” GP 17
	“I’ve essentially look at the GLA:D® providers in their local suburb and that’s how I’ve done it.” GP 19
EMR template: Only one GP reported using the EMR template.	“I downloaded the template… having the best practice software or some templates that that's useful… I haven't used it, but I could use it.” GP 26
	“I didn't know that [EMR template] would integrate with my software…so I looked at the template and I thought oh I’ve got to fill in this and then I'm gonna write my letter.” GP 11
Web-based toolkit: A few GPs stated they used the web-based toolkit. e GPs reported struggling to remember to use the web-based toolkit, with some preferring printed handouts and not having to go to another website.	“No, sadly, I don't think I have [used the online toolkit]...I'm just bad at remembering to do these things.” GP 3
	“[Online toolkit has] been pretty useful because, I mean when you do the printouts and then you hand it to them, they sort of realise what they need to do in terms of getting better.” GP 11
GLAD® Posters and flyers: Some GPs said the posters were helpful to instigate conversations about exercise, but others didn’t use them.	“The posters – these are the visual things that patients take most out of. And a lot of people who have never seen me, they’ve just seen the posters in the main waiting area, they would say, “Okay, I want to talk about that.” GP 19
	“Our practice owner she doesn't like anything hanging on walls anywhere. So, personally, I wouldn't put anything on the wall. GP 26
Booster session: GPs described the booster sessions as useful to keep GLA:D® top of mind and to be connected with local GLA:D® clinicians.	“I probably need to use more of the local providers…it was really helpful to catch up with somebody locally today, that’s good.” GP 19
	“I think the follow-up [booster] has meant just reminded me each time that I'm like, “Yeah, that's right, I've learned about this.” It's just kept it in my head as something to offer.” GP 3
Physiotherapist communication template: No GP mentioned receiving the REFER communication template from physiotherapists, although two GPs mentioned receiving letters. Almost all GPs stated they would appreciate this communication, especially if it included patients function pre/post GLA:D® program.	“That would be perfect to have a pre-and-post kind of exercise to see how we have travelled and I suppose that gives the patient also a great emotional advantage to know that, “Yeah, this works and I know what I need to be doing”…this is the evidence that you’ve done it well.” It’s definitely gonna be helpful.” GP 19
	“Yep. [The physio] wrote a letter back after – it was a few months, to say how they were going and their progress… it's useful of how people are progressing. So I really appreciate it absolutely, very, very much so.” GP 27
Adaptations to REFER components: The main adaptation made to REFER was in how GPs preferred to refer patients to GLA:D: either by typed letter, hand written note, or by sharing the online GLA:D® map with patients to facilitate self-referral.	“I've got [the EMR template] in in my system, but no, I actually tell patients to self-refer because it puts their responsibility onto them and then they can then do their research, who they want to go to and then take themselves along to whoever they chose.” GP 17
	“I find out if they’re doing the GLA:D® program and I’ll write it down a piece of paper for them...I still use a script on a piece of paper. It’s the fastest thing for me before I can open anything on the computer…So, it all seems so hard that I end up just handwriting it.” GP 5
Maintenance: Sustainability and scale-up of REFER
THEME	SUPPORTIVE QUOTATIONS
Future Intentions of using REFER: Some GPs spoke positively about intending to use the web-based toolkit and posters in the future. One mentioned intending to trial the EMR template.	“Because I've only had one real [osteoarthritis] patient recently, I probably haven't done much, but I think [the online toolkit] will be useful in the future for me to go and look at.” GP 3
	“I'll print out the posters and I'll use that and I'll probably use that when I'm talking to people about where they want to go, especially that initial consult. And I'll use the referral templates as well. I'll download that as well.” GP 27
Future improvements to REFER: GPs main suggestions were around improving the functionality of the online GLA:D® map and better facilitating connections with local GLA:D® physiotherapists. One GP also wanted more detailed information about the GLA:D® program. Another GP suggested interventions need to target patients on value of exercise.	“If you actually just meet the person who's doing [GLAD®] and you know them, I think you're more likely to refer to them, because they come to your mind when you're seeing somebody, you are like, “Oh yeah, I remember that so and so was doing this program. That sounds like a good idea. I'll refer to them.” So I think that you would have more referrals if you did that.” GP 27
	“I think I thought it was going to be more, “What is the actual program?” and we did get a little bit of that with the physios but I was interested in what – but that's probably more from a physio perspective, even than a GP perspective.” GP 10
Strategies for increasing future GP engagement with REFER: Included targeting GP registrars, promoting the study through RCAGP or newsletters, targeting practice managers, nurses, and other practice staff, offering to do GLA:D presentations, and continuing to offer financial incentives and CPD points for GP participation.	“You guys did offer a financial incentive..so, if that is happening make it a big publicised because then people are more inclined to do something if there's a bit of a token to it, but I think the big seller is whether it can be turned into RACGP's CPD points… because they’re getting something out of it.” GP 28
	“Having a meeting with a number of GPs and maybe doing a presentation or something would be good.” GP 11
	“You need testimonials from the surgeons, and from GPs and from physiotherapists and from patients.” GP 1
	“Getting some coverage in GP media...So even being “new funded program through community health centres, the GLA:D® program”, or “evidence-based OA program now funded through the community health centres”, and then using that as a way to, one, talk about the GLA:D® program.” GP 1
	“Advertising it for the GP registrars. Yeah, I think that would be amazing too. One of the worst things when you’re a baby GP is musculoskeletal medicine.” GP 5
	“Trying to write papers that end up in places where GPs read.” GP 1
	“You can advertise the program on RACGP. That would be an easy way where people would find it.” GP 1

GP general practitioner, GLA:D®, Good Life with osteoArthritis in Denmark, EMR electronic medical records, RACGP Royal Australian College of General Practitioners

REACH

The intervention reach extended to GPs with diverse motivations for enrolling in the study and whose practices were highly variable with regards to patients’attitudes and beliefs about exercise, level of health literacy, and ability to afford physiotherapy.

Effectiveness

REFER was considered effective, with most stating that after the workshop they were more confident and knowledgeable in managing OA, better understood the GLA:D^® program (including how/when to refer), and were referring more patients to it.

Adoption

All GPs, except one, reported the intervention to be highly acceptable and spoke positively about the workshops, both live online and self-directed delivery modes.

Implementation

The reported use of other intervention components beyond the workshop was low, with the most used aspect being the online GLA:D^® location map. The main adaptation made by GPs was to refer patients to GLA: D^® using a typed letter or their own EMR template. Alternatively, some GPs preferred to give patients a handwritten note with details about the GLA:D^® website or GLA:D^® locations to facilitate a self-referral. Facilitators for implementation of REFER included it being flexible, having a pre-existing interest in OA, and having a caseload of patients who could afford to attend the GLA:D^® program. Barriers to implementation included lack of time, managing complex public funding schemes, patients have multiple health issues or being unable to afford to attend GLA:D^®, and lack of communication from local physiotherapists when they did refer a patient.

Maintainance

GPs suggested numerous strategies for expanding GP recruitment in the future including targeting GP registrar training programs or practice managers, promoting through RACGP, and continuing to offer incentives (e.g. CPD points, financial token).

Mixed methods data synthesis

An overall summary of the mixed-methods REFER evaluation findings is displayed in Table 6.

Table 6.

Joint display summarizing the mixed methods intervention evaluation guided by RE-AIM QUEST framework

RE-AIM	QUAL finding	QUANT finding
Reach: Understanding GPs who participated in the intervention	• GP motivations for participating in REFER included to increase OA and GLA: D® program knowledge, obtain CPD points, receive financial incentive, and to be involved in research. • GPs practices were highly variable regarding frequency and socioeconomic status, health literacy, and health or exercise beliefs of patients with OA.	• 27 GPs (GPs were recruited from 4.3% of GP practices contacted) • 52% women • Mean age = 47 years (SD = 10) • Clinical experience: 19% < 5 years; 56% 5–20 years; 26% >20 years • 33% previous OA professional development training • Patients with OA seen weekly: 44% <5; 37% 6–15; 19% 15 or more • Median 20% (range = 0, 68%) of patients with OA currently referred to orthopaedic surgeon (pre-intervention) • Believe an OA treatment is effective based on patient*: satisfaction 56%; self-reported pain or QoL 85%; function 63% (pre-intervention)
Effectiveness: Impact of REFER on GPs’ OA knowledge, confidence, and referral behaviour	• Almost all GPs reported feeling more confident and knowledgeable in managing knee OA, including when and how to refer to the GLA: D® program after completing the workshop. • Most GPs reported that they had changed their behaviour to now refer patients to GLA: D®, whereas GPs who were already referring patients to GLA: D® stated GLA: D® referral was now more ‘top of mind.’	• Compared to the pre-workshop, at post-workshop: • A greater proportion of GPs’ correctly answered knowledge questions related to exercise not being a risk factor for OA (82% vs. 96%); exercise, education, and weight management as first line treatments (78% vs. 92%); benefits of exercise regardless of pain or radiographic changes (78% vs. 96%); and RACGP guideline recommendations for use of pain medication [conditional recommendations for NSAIDS (11% vs. 46%) and strongly recommended against opioids (52% vs. 77%)]. • GPs’ confidence and beliefs in OA management significantly improved in 63% of questions related to being knowledgeable about RACGP guidelines, feeling as though they have had training in OA management, knowing availability of and how to refer to physiotherapy-led services, and understanding the purpose and role of OAHKS. • A greater proportion of GPs’ believed exercise-therapy and education was very effective.
Adoption: Acceptability and satisfaction with REFER intervention	• All GPs, but one, reported REFER to be highly acceptable across all acceptability domains.	• 73% and 27% of GPs ‘strongly agreed’ or ‘agreed’ that they were satisfied with the workshop
Implementation: Fidelity to use of REFER intervention components	• GPs reported a diverse range of facilitators (e.g. pre-existing relationships with local GLA: D® physiotherapists) and barriers (e.g. patient beliefs, costs) to implementing REFER. • GPs all attended and highly valued the workshop. Most also used the online GLA: D® map, but implementation of other intervention components was low.	• 54% and 38% of GPs ‘strongly agreed’ or ‘agreed’ that attending the workshop will change the way they diagnose patients with OA • 65% and 27% of GPs ‘strongly agreed’ or ‘agreed’ that attending the workshop will change the way they manage patients with OA
Maintenance: Sustainability and scale-up of REFER	• Some GPs spoke positively about using aspects of the intervention going forward and made suggestions for minor improvements in the future. • GPs suggested ways for scaling up GP involvement in the future including targeting recruitment, where to advertise, and continuing to offer incentives.	NA

GPs general practitioners, GLAD^® Good life with osteoArthritis in Denmark, CPD continuing professional development, SD standard deviation, RACGP Royal Australian College of General Practitioners, OAHKS Osteoarthritis hip and knee service

*Could select multiple options

Discussion

This mixed-methods study sought to co-develop and evaluate the feasibility of the REFER intervention to increase GP referrals of people with knee OA to an osteoarthritis management program. GP recruitment was challenging, resulting in limited reach of REFER. Yet, GPs who engaged with REFER viewed it as highly acceptable. Workshop attendance was associated with improved GP knowledge, confidence, and beliefs related to OA management and guidelines as well as greater awareness of community-based referral options for first-line care including GLA:D. Before scaling up this intervention, considerations should include new recruitment strategies, creating a more individualized GP interventions, and exploring how to better measure GP referrals (e.g. chart audit). These considerations would be prudent for other researchers to reflect on in the development of future interventions that seek to change GP exercise referrals with regards to management of other chronic conditions, where the uptake of exercise services is low, such as cardiovascular disease [45] and diabetes [46].

Reach of REFER to engage GPs was limited, with less than 1 in 20 GP clinics contacted having a GP who agreed to participate. Multiple communications were required to arrange and confirm participation once consent was obtained, despite GPs being compensated for their time for both interviews and workshop attendance. This challenge to recruit GPs is consistent with other published studies seeking to implement and evaluate primary care-based models of OA care [25, 47]. As primary care providers GPs are tasked with the diagnosis and management of many chronic conditions, of which OA is only one. Our qualitative findings provide key insights into why GP recruitment is so challenging. In addition to the recruitment challenges articulated by VicREN, our GPs expressed key barriers to professional development and engagement in research as the associated costs, lack of time, and competing priorities. These findings align with challenges articulated by others attempting GP recruitment in Australia [25]. These experiences reinforce that developing and evaluating strategies to increase GP engagement in research should be a priority. GPs participating in our process evaluation interviews offered additional suggestions to enhance recruitment which could be tested in future trials. This includes focusing on GP registrar training programs, targeting GP practice managers or nurses, or having physiotherapists reach out to GPs. Most GPs in our study stated they valued communication from physiotherapists about patient outcomes, and that having relationship with a local clinician was seen as a facilitator to referral. Previous work has also highlighted the importance of increasing interprofessional communication to improve OA care [48]. This also fits with previous research that has identified that friends or acquaintances are key facilitators for GP engagement in research [49].

GPs who did engage with REFER considered it highly acceptable, including its usefulness, low burden, alignment with their values, and ease in understanding education content. All GPs participating in this study were satisfied with the workshop component of REFER. Strong acceptability and satisfaction with REFER may be a function of our efforts to meet the needs of GPs we interviewed in phase 1. Specifically, we minimized barriers to access by making our intervention freely available and providing the flexibility of an online workshop, offered live or self-directed. We also offered a financial incentive and supported GPs to self-log CPD points. GPs interviewed in phase 2 also spoke very favourably of the web-based GLA:D clinic location map. Addressing suggestions to improve the functionality of this map (e.g. adding details about cost, times, languages catered for at each GLA:D location) may help to further enhance the acceptability and effectiveness of REFER in the future.

Based on input from GPs in phase 1, all intervention components supporting the workshop were designed to be low burden and keep GPs’ time requirements low. Each was discussed and broadly endorsed by participants in phase 1. Yet, they were not strongly embraced by GPs in this study. Numerous other studies working with GPs have also reported low levels of intervention adherence illustrating this is not a unique problem [25, 49]. Notably, the EMR template was not used, which meant we were unable to quantitatively measure GP referrals to the community-based first-line care program. Reflecting previous studies which have successfully implemented e-templates with GPs in OA care [50, 51], providing greater in-person assistance with installation and operationalization of the REFER template may improve adoption in the future. Some GPs in our study also expressed they preferred handwritten referrals or guiding the patient in self-referral to facilitate patient autonomy. Contrary to this, other GPs suggested making REFER more prescriptive with less choice might improve GP compliance. This conflicting feedback highlights diverse GP preferences in referral behaviours and may reflect need to individualize referral behaviour change interventions to each GP’s preferences.

The high acceptability of the REFER workshops among GPs who did engage may help to explain its associations with improved knowledge, confidence, and awareness of referral options for first-line care. These results in our small sample of GPs suggests that REFER could be scalable across Australia if we first understand how to better recruit and engage with GPs. Given the evidence that patients are more likely to engage with physiotherapy when referred by a medical specialists (including GPs) [20, 52], this work should be prioritized to improve guideline-based care for patients with OA.

This feasibility study has numerous strengths that align strongly with key recommendations to improve primary care-based models of OA care [26]. Development of REFER was grounded within a recognized behaviour change conceptual framework aligning with implementation theory (e.g. BCW) [35], and involved key stakeholders (e.g. GPs) early [26]. Our mixed methods evaluation, guided by the RE-AIM QuEST framework [29], is one of the few studies to acquire provider-level outcomes from multiple data sources (pre/post surveys, interviews). Our behaviour change intervention was described as highly acceptable, associated with improvements in knowledge, confidence, and self-reported referral behaviours, and the online nature of the workshop and GLA:D location map makes REFER potentially highly scalable.

Some limitations should be considered in the interpretation of this feasibility study. Given the extensive recruitment efforts and the low proportion of GPs who agreed to participate, participants in this study may represent a unique sub-set that are more interested in OA management than the average GP. This may have been reflected the GPs’ pre-workshop beliefs about the high effectiveness of different OA treatments, including exercise, education, and weight management. These beliefs about effectiveness may be different in other GPs. The GLA:D^® program is one example of an osteoarthritis management program, and our findings may not be generalizable to other programs, physiotherapy and/or exercise services.

The EMR template was not used by GPs which meant we had no quantitative data directly measuring referrals. Further, while REFER focused on GP-related barriers to accessing evidence-based first-line care for people with knee OA, other significant barriers remain. These factors may be best conceptualized using the socioecological model [53]. At a patient level, barriers such as out of pocket costs, lack of transportation, beliefs and misconceptions about exercise making their OA worse, and conflicting work or family commitments are reported to influence access [19, 20, 52]. At a systems level, barriers include a lack of access to publicly subsidized allied health services, poorly integrated interdisciplinary services, long wait times, and complex referral pathways [20, 21, 54, 55].

In conclusion, this project set out to co-develop and evaluate REFER, a multi-faceted, theory-based, behaviour change intervention to increase GP referral of people with knee and hip OA to an osteoarthritis management program. Our findings indicate REFER may have potential to improve referrals to community-based osteoarthritis management programs. REFER had high acceptability and was associated with improved GP knowledge, confidence, and beliefs pertaining to OA management, guidelines, and referral options. Importantly, this work may provide guidance for future studies planning GP targeted interventions. Prior to scale-up, work is needed to improve reach and engagement with GPs, and to further refine the intervention.

Authors’ contributions

AE, CB, and JMN conceptualized the study. AE collected and analysed all data with contributions from AG, DD, MP, JK, and CB. AE drafted the manuscript, then all authors read, provided input, and approved the final manuscript.

Funding

This project was supported by an Internal Investment Schemes Early Career Researcher grant (2021) from La Trobe University.

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to partticipate

This study received ethics approval from La Trobe University (HEC21424) and this study adheres to the declaration of Helsinki. All participants provided written and verbal informed consent.

Consent for publication

Not applicable.

Competing interests

AME and both CJB and JK are program leads of GLA:D Canada and GLA:D Australia, respectively. GLA:D is an international, non-profit implementation initiative aimed at implementing clinical guidelines for osteoarthritis in practice. Otherwise, the authors declare they have no competing interests.