General practitioners’ views on managing knee osteoarthritis: a thematic analysis of factors influencing clinical practice guideline implementation in primary care

Osteoarthritis (OA) is a highly prevalent, disabling condition ranked the eleventh highest contributor to global disability [1, 2]. Knee OA is mostly diagnosed and managed in family practice settings [3] and principally by general practitioners (GPs, i.e. family doctors) [4]. Clinical practice guidelines (CPGs) recommend diagnosing OA using clinical criteria without imaging, and facilitating self-management through education and the provision of advice on weight management and increasing physical activity [5]. However, care inconsistent with these recommendations has been identified in several countries and health care settings [6–11]. GPs tend to over-rely on imaging for diagnosing OA [12, 13], under-emphasise exercise and weight loss options in favour of drug and surgical management [6, 9, 14], frequently refer for ineffective arthroscopic surgery [15], or refer for joint replacement surgery before an adequate trial of recommended conservative treatments [10, 14].

There is a need to develop effective strategies that facilitate GPs’ uptake of recommended OA management practice. Detailed behavioural analysis of the reasons behind the inadequate uptake will help inform implementation interventions [16]. There have been previous qualitative studies asking primary care practitioners about topics related to the provision of care for people with knee osteoarthritis [7, 17, 18] and osteoarthritis more generally [19–21]. Common findings include trivialising or normalising the problem, lack of knowledge/skills, and resource issues [22]. However, previous studies on the topic have not focussed in any depth on the barriers and facilitators to implementation of the priority recommendations currently identified as being the most underutilised in care globally [22]. The aim of this study was therefore to identify barriers and facilitators influencing whether GPs perform the activities of: 1) making a clinical diagnosis without imaging, 2) engaging patients in exercise and physical activity, and 3) engaging patients in weight loss. The study involved the systematic and comprehensive identification of behavioural drivers related to providing this care for people with knee osteoarthritis with the hope of uncovering new and useful additional findings. We used a novel framework to guide our classification and labelling of themes, and included a discussion of the results in the context of previous findings.

This study is part of a larger project (PARTNER) to increase delivery of recommended knee OA management within Australian primary health care. All GPs provided informed consent to be interviewed and recorded. The reporting of this study adheres to the COnsolidated criteria for REporting Qualitative studies (COREQ) 32-item checklist [23].

Using a theoretical model of behaviour, this study identified eleven key drivers of GP behaviour that impact implementation of recommended practice for diagnosis and non-drug, conservative management of knee OA in the Australian primary care setting.

Our findings suggest whilst GPs mostly know that knee OA can be diagnosed without imaging and that exercise and weight loss are recommended, their described behaviours in practice were often discordant with their own knowledge. We identified several barriers that lead to this discordance. Despite participating GPs describing OA diagnosis and management as straightforward, comments indicated incomplete or inaccurate knowledge of OA disease processes and prognosis, and low confidence with diagnosing knee OA without x-ray. Knowledge that goes beyond the general guideline recommendation to the what and why of exercise and weight loss interventions, and confidence in their ability to facilitate these interventions effectively, also appeared insufficient. Most, but not all, GPs were aware of their lack of knowledge and confidence. Feelings of being ill prepared to manage OA are consistent with other studies [7, 18, 22, 30, 31]. GPs in our study tended to adopt a simplistic model of OA which neglects the involvement of joint structures other than cartilage, promotes a biomedical model of the disease and its consequences, and may contribute to a fatalistic attitude among patients. Understanding knee OA as a problem of chronic pain involving broader psychosocial factors and as a condition that affects multiple joint structures [32] and therefore not requiring imaging for diagnosis [29], is likely to require a substantial shift in long-held beliefs and habits for many GPs.

GPs’ reflections on conversations with their patients about OA identified a reliance on terms that normalise OA (e.g. OA is to be expected with aging), and a preference for vague, general terminology such as ‘wear and tear’ to communicate the diagnosis to patients. Previous research suggests a dissonance between GPs’ rationale for avoiding articulating the diagnosis and how a vague diagnosis is perceived by patients. Clinicians play down an OA diagnosis in an attempt to facilitate acceptance and avoid upsetting patients [30]. However, using dismissive or reassuring terms over factual explanations and empathy can be interpreted by patients as the GP trivialising their problem [18, 30, 33] and lacking interest in their debilitating symptoms [30, 34]. GPs may therefore unintentionally be acting as a barrier to patients making beneficial lifestyle changes, communicating in a more top-down rather than collaborative way with patients about OA diagnosis and treatment options.

Time pressure was identified in our findings and has been widely cited previously as a barrier to GPs ability to implement CPG recommendations including facilitating lifestyle changes [22, 30, 31, 33–36]. The problem is exacerbated when patients with OA have multiple comorbidities and GP consultation length averages only 14–15 min [37]. A UK study found limited time (13 min) and the presence of multiple co-morbidities (3 conditions on average) led to GPs spending minimal time on OA management and prioritising other conditions they perceived as greater threats to patient health [33]. However, OA imposes a substantial burden on individuals and impacts health-related quality of life to at least the same degree as other common chronic diseases [38, 39]. GPs placing less importance on OA management than the patients themselves may lead to under-management in primary care and patients feeling unsupported and dissatisfied [40].

While some of the barriers to evidence-based management found in this study may be specific to managing OA, many of the barriers have also been found in previous qualitative and quantitative (survey) studies on other chronic conditions including, for example, low back pain [41–43], diabetes [44, 45], chronic kidney disease [46, 47], chronic pain [48], depression [49] and obesity [50, 51]. Frequently occurring barriers include incongruency between patient wishes and guideline recommendations [41–43, 48, 49], suboptimal practitioner skills for patient education [42, 45–47, 50] – in particular, the difficulty communicating a non-biomedical explanation of a biopsychosocial problem [41, 43, 48], difficulty ‘selling’ lifestyle change and providing support for better self-management [42, 45–48, 50, 51], frustration with patients [45, 47, 50, 51], lack of time [41, 43, 45–49, 51], limited access to other services to help with management [41, 42, 46–49], and resistance to changing practice habits [44, 49]. Collectively, these problems may reflect the challenges in managing chronic conditions within a system designed primarily to manage acute illnesses [52, 53]. On the plus side, the commonality across many conditions, especially around lacking skills and confidence in having conversations about obesity and supporting patients to increase physical activity, means that addressing some of the barriers to optimal OA care will be transferable to improving management of many other conditions and vice versa, particularly given many patients have multi-morbidity.

Our findings identified multiple interacting barriers influence GPs’ implementation of OA guideline recommendations, suggesting complex, multimodal solutions may be required. Targeted GP education and training interventions to build motivation and confidence were potential facilitators to clinical guideline adherence identified by the GPs in ours and other studies [54, 55]. Changes to the guidelines themselves may be beneficial. Currently, OA guidelines lack specific exercise and weight management recommendations and are open to variable interpretation [56, 57] potentially resulting in GPs feeling ill equipped to deliver lifestyle interventions [22]. Further research to identify optimal exercise types and dosage, and effective weight loss interventions is required, however it is currently feasible to suggest specific exercise programs based on existing exercise science and general exercise and physical activity recommendations [58, 59], and to provide guidance on how to have effective conversations with patients to facilitate adoption of lifestyle change recommendations based on principles of patient-centred care and health behaviour change [29, 60]. Guidelines could also provide clear, plain-language statements that can be readily used by GPs during consultations to help with OA management discussions [61]. This type of communication guidance has been demonstrated to facilitate GP uptake of CPG recommendations [34]. Finally, incentivisation and/or coercion-based interventions may help address motivational barriers. This is based on the work of Michie et al. who developed the Behaviour Change Wheel [16], which is an evidence-based framework for planning behaviour change interventions [24]. The framework provides a systematic and theoretically guided method for identifying the types of interventions that could be expected to be effective. Incentivisation and coercion are behaviour change intervention ‘functions’ that are suggested for addressing some of the types of barriers we found among our GP sample, namely reflective and automatic motivation barriers [16]. Incentivisation has previously been used to drive GP adoption of patient-centred care; however, results have been varied [62]. Costly pay-for-performance interventions appear to influence short-term GP behaviour change, but not long-term, and do not appear to translate to improved patient outcomes [63, 64]. Coercion in this context means creating an expectation of cost. Costs could include financial loss or negative feelings about the undesired behaviour(s). Thus, examples of coercion in our context could include reducing rebates for care consistent with the undesired behaviour, providing education about the negative consequences to patients when sub-optimal care is provided, or portraying those behaviours as ‘old fashioned’. Michie et al. [16] note that whether these functions are effective or not depends on the behaviour and the circumstances and should be thought of as options for consideration. Coercive interventions seem not to have been investigated as yet, and it may be that they are not perceived to be acceptable in this context. Utilising social influences via communities of practice or local opinion leaders, or diffusion of innovations aimed at shifting practice behaviours may be better options for addressing reflective and automatic motivation barriers [65].

Several strengths and limitations should be considered when interpreting the findings beyond the context of this study. The attitudes and beliefs of GPs willing to participate in research may not be representative of all GPs. Most GPs interviewed had at least 20 years experience and the views and practices may differ amongst GPs with less experience and/or those who completed training more recently. These, and other location-related contextual factors, including system-related findings pertaining specifically to the Australian healthcare system, should be considered when transferring findings to other contexts. However, it is likely that many findings will be relevant across contexts. In addition, the data represent only the version of GP’s perceived reality they wanted to share with the interviewer, and interpretation is influenced by the analysis team. Thus, there may be important factors that influence GP practice behaviour not detected by the study.

In summary, our analysis of Australian GPs’ discussion of implementing core underutilised CPG recommendations for knee OA management identified multiple influences that impact practice behaviour. Key negative influences identified were knowledge gaps, low confidence and skill deficiencies, time and other system constraints, and the GPs’ perceived role, assumptions about patients and established habits. Positive influences include the benefits of healthy lifestyle changes for all patients, GP optimism and using a patient-centred approach. The complexity of these influences suggests complex, multi-model solutions may be necessary including changes to clinical guidelines, targeted education and training, the implementation of new models of service delivery and exploitation of positive social influences. Such interventions may help bridge the evidence-to-practice gap, which is almost certainly needed if the individual and societal burden of knee OA is to be reduced.