Search strategy

Searches were undertaken without language restrictions for both published and grey literature that considered models of service delivery and care pathways. All sources were searched for evidence published between 1 January 1990 and 20 December 2017. The following databases were searched: Medline, Medline in Process, Cumulative Index of Nursing and Allied Health Literature, Embase, Pedro, PubMed, Web of Science, Cochrane Central Register of Controlled Trials and Health Management Information Consortium. In addition, websites were searched for grey literature, care pathways and policy documents. Finally, reference lists were checked and citations tracked for included evidence. Full details of the search, including the Medline search strategy are provided in the published protocol.21 Details of searches for other databases are available on request from the authors.

Eligibility

All study designs and documents that considered care pathways for musculoskeletal hip and/or knee pain for adults from the point of referral for specialist opinion meeting the European Pathway Association definition of a care pathway22 were included. No diagnostic criteria were specified as musculoskeletal pathways accept all hip/knee pain without such criteria. Publications were excluded if their primary focus was generalised inflammatory arthropathy, stroke, postsurgical care or musculoskeletal care pathways that did not include hip and knee joint-specific data.

Data selection

After duplicate results were removed, two reviewers (KB and Cheryl Cleary, FM) independently considered citations against inclusion/exclusion criteria. At both title/abstract and full-text stages, disagreement was resolved by discussion with a third reviewer.

Quality assessment

Studies were appraised using validated checklists for specific research designs: The Critical Appraisal Skills Programme checklist for randomised controlled trials (RCTs)23 and the Joanna Briggs Institute checklists for qualitative studies, cross-sectional surveys, longitudinal studies and case series.24 After a sample of studies had been piloted to ensure a baseline of understanding and agreement between reviewers, papers were assessed by one reviewer (KB) and checked by a second (FM).

Data extraction and synthesis

Data were extracted into a standardised form. After piloting, data extraction was completed by one reviewer (KB) and checked for accuracy by another (FM). The protocol21 specified the utilisation of a mixed-methods approach to data synthesis. However, only one qualitative study was identified. This combined with the low quality and heterogeneity of the evidence precluded any mixed-methods synthesis or meta-analysis. Consequently, the results are presented narratively by primary outcome type reported. These categories were identified inductively from the data: patient flow (the movement of patients between healthcare professionals, services and organisations),25 professional competency (the level of agreement in diagnosis and treatment between Advanced Physiotherapy Practitioners and orthopaedic surgeon or expert physicians), resource use including clinical and cost effectiveness, time (changes in patient waiting times experienced as a direct result of the way the pathway was delivered), patient satisfaction and patient experience. These outcomes are organised according to where they apply in a multilevel, whole systems approach, that is, macro, meso or micro level.6

Patient and public involvement

A patient representative was involved at the planning stages and helped to formulate the aims of this systematic review.

Meso level: patient flow

Johnson et al 29 evaluated an orthopaedic-led fast-tracking pathway for hip replacement. Physiotherapists assessed patients on the orthopaedic waiting list that an orthopaedic surgeon had screened off for physiotherapist management. The fast-track pathway resulted in 23/25 patients being given a total hip replacement, but a further 15 patients from the routine route were also listed for joint replacement surgery.

A study by Pearse et al 30 benchmarked the ability of advanced practice physiotherapists to independently assess 85% of patients on an orthopaedic waiting list, following strict criteria set out by a surgeon. Fifty patients with knee pain had been screened by a surgeon and deemed appropriate for physiotherapy assessment and non-surgical management. Physiotherapists failed to meet the benchmark; only independently assessing 66% of patients. Of the patients they were not able to manage, 17 (34%) were referred to a consultant, with only 11 being listed for surgery. The authors concluded that physiotherapists did not efficiently manage patients on the orthopaedic waiting list. However, the benchmark set was not underpinned by research.

The Damask Study Group26 41 42 conducted a RCT to assess the effectiveness and cost effectiveness of GP referral to MRI and provisional orthopaedic appointment, compared with referral to an orthopaedic specialist only, for knee problems. The study found that having an MRI did not change the diagnosis or the treatment plan, but there was greater treatment and diagnostic confidence among GPs in the MRI group. Patients in this group also reported a statistically significant improvement in their physical functioning and quality of life.

Gwynne-Jones et al 15 evaluated the effectiveness of a prioritisation service for assessment of hip and knee OA. A physiotherapist and a nurse assessed patients referred into a joint clinic and surgeons only assessed those identified at referral as requiring specialist opinion, or who were subsequently rereferred from the joint clinic. Of the 339 patients included, 54 were referred directly to the surgeon and a further 89 were referred after the initial assessment. Of the 143 referred to the specialist clinic, 115 were listed for or underwent surgery. The authors concluded that this service freed up valuable surgeon time.

Only one study by Inglis et al 38 reported outcomes that can be applied at a macro level within a complex multilevel system. This evaluated the effectiveness of consultant-led triage of referrals for patients requiring a fast-track specialist surgical opinion. Of the patients referred, 57% of those with hip pain and 46% of those with knee pain were assessed or directly listed. Others were sent elsewhere because of insufficient clinic capacity, or they were deemed lower prioritisation based on referral information, or their referral was incomplete.

Meso level: professional competency

Overall, agreement in diagnosis between the professional groups was reported to be high for three studies, with kappa values ranging from 0.80 to 89 agreement,33 35 37 where 1.0 represents perfect agreement. The study by MacKay et al 36 reported agreement between physiotherapists and orthopaedic surgeons in 69% of cases. Compared with arthroscopy, physiotherapists were found to identify the correct diagnosis in 80%–84% of cases.27 Agreement on treatment type (surgical vs non-surgical) between physiotherapists and physicians was found to be 77% in one study35 and Decary et al 37 reported a kappa value of 0.73.

Macro level: resource use

The implementation of a stepped care pathway in primary care was described by Smink et al.40 The pathway resulted in the increased use of non-surgical treatments for hip and knee OA over 2 years. Dietary interventions were less well used. Of the patients 21% were referred for orthopaedic opinion in the first 6 months and at 2 years, this had risen to 45%.

Meso level: resource use

A number of studies described the resources used at the point of referral for a specialist opinion but there were differences in the type or structure of the pathway being reported on.

Farrar et al 28 evaluated a multidisciplinary team assessment clinic (MCAS), compared with direct referral to an orthopaedic service. Referral to the MCAS clinic resulted in more appointments before a working diagnosis was achieved compared with the orthopaedic route (p<0.001). However, there was a greater use of non-surgical treatments and lower surgical referrals in the MCAS service (16% vs 36%; p<0.001).

In a study by Rabey et al, 31 9% of new referrals seen by an Advanced Physiotherapy Practitioner were referred for surgical opinion. Of these, 42% were for knee conditions. Of the patients referred with knee conditions, 84% went on to have surgery.

Only one study16 described a potential cost saving by using advanced physiotherapy practitioners to directly list patients for total hip replacement surgery rather than going via an orthopaedic clinic. This saving was calculated to be £145 per patient. MacKay et al 36 found that for ongoing treatment in a clinic, physiotherapists were more likely to give exercise and education at the time of consultation while a surgeon was more likely to refer to rehabilitation services (which would include exercise and education), adding a further step and potential delay in the patient receiving the correct care.

Desmeules et al 35 did not find any difference in the referral rates for imaging between orthopaedic surgeons and Advanced Practitioners. Conversely, an MCAS service was reported to result in a greater use of imaging (p=0.04).28 In terms of numbers of imaging referrals made by Advanced Physiotherapy Practitioners, Rabey et al 31 reported that 36% were referred for a knee X-ray and 23% for an MRI.

Meso level: time

Physiotherapy-led triage resulted in a reduced wait for surgery from 140 to 40 days.34 Using physiotherapists to directly list patients for total hip replacement, rather than going via an orthopaedic appointment resulted in a shorter wait for surgery (21.4 weeks vs 24.7 weeks).16 Conversely an MCAS service resulted in a longer waiting time from referral to diagnosis (p=0.05).28 Doerr et al 39 described a new referral process and found a reduced patient wait for an initial specialist appointment but the time is not specified. Desmeules et al 35 found that advanced physiotherapist practitioners had longer consultation times than orthopaedic surgeons.

Micro level: patient satisfaction and patient experience

Two studies34 35 evaluated patient satisfaction using questionnaires. Both found a high level of satisfaction with the physiotherapy triage. One35 reported greater patient satisfaction with the care by the physiotherapists than the surgeon and the other34 reported that no patients requested a subsequent appointment with the orthopaedic surgeon.

Only one interview study evaluated patient experience.32 This related to being on a waiting list for specialist diagnosis and surgery. Four themes were identified; inadequate information, the social and psychosocial cost of waiting, coping strategies and the varying effectiveness and support provided by clinicians. The recommendation was to improve information provision to improve patient well-being.

Macro level

The findings from only one study were relevant at the level of healthcare systems and policy. Smink et al,40 described healthcare utilisation after the implementation of a stepped care strategy in the treatment of hip and knee OA. The authors reported that the pathway resulted in a higher percentage of patients being managed with non-surgical treatment before referral for specialist opinion, but that referral to a dietician for weight management was underutilised. However, the usefulness of this research is limited by the lack of baseline data to compare healthcare utilisation premodel and postmodel implementation. The acceptability of this model to GPs and patients also needs to be questioned. While 157 GPs were recruited from eight practices, healthcare utilisation data were only available for 313 patients. Two patients per GP is a very low number considering how many individuals present with joint pain. Second, patient experience of the pathway does not appear to have been considered. Further research on the effectiveness of the pathway for GPs and patients, along with greater integration of dietician referrals should provide additional answers.

Meso level

Most of the studies included in this systematic review looked at patient flow (fast-track models, referral prioritisation and advanced practice triage) and resource use. The findings suggest that advanced practice physiotherapists are comparable to orthopaedic surgeons in their diagnosis of knee and hip pain27 33–37 and triage to surgical or non-surgical treatment. This would suggest that physiotherapists have the skills required to triage patients referred for a specialist opinion about their hip or knee pain.

Despite this finding, there appears to be less consistency in treatment allocation and healthcare resource use when applied in practice across different care pathways. This may be due to differences in staff skill mix, triaging processes and patient referrals. For example, in one study higher conversion rates to surgery are reported from an orthopaedic clinic, while advanced physiotherapy practitioners have higher referral rates to non-surgical treatment options.28 In addition, practice within orthopaedic clinics can vary with physiotherapists providing more exercise and advice and surgeons having higher referral rates to rehabilitation services.36 Surprisingly, considering the importance of weight management in the treatment of joint pain and OA, no studies have specifically evaluated referral rates to dietician services by these different professional groups.9 44 45

In terms of imaging referrals (X-ray and MRI), there is inconsistency in the referral rates across different clinics or professional groups.28 35 A RCT evaluating the clinical and cost effectiveness of GP referral for specialist opinion26 41 42 found that clinicians had more diagnostic and treatment confidence when referring for an MRI and that patients on this pathway had a better level of functioning. The concern about the study is that it only considered GPs as the primary referrers. It also failed to take into consideration other treatments recommended within the primary care pathway.9

From the evidence, it is clear that, regardless of nuances in the care pathways considered, there will be a proportion of patients assessed in a physiotherapy clinic who will require a surgical opinion and also patients referred to an orthopaedic clinic who will require conservative management.15 16 30 38 This occurred despite triaging and referral systems implemented to fast track or direct patients to the most appropriate service.15 29 38 Additionally, both the fast-tracking criteria29 and surgeon screening of referrals,30 seem to lack sensitivity and specificity as individuals move between surgical and conservative pathways. The quality of the research means that insufficient evidence exists regarding predictor variables to identify the key variables to inform specialist referral. Thus, the current evidence suggests that an innovative workforce model is required that integrates the skills of the different professionals to meet individual patient needs and tackle these complexities. This would give patients access to clinicians with expert knowledge of both conservative and surgical treatments at the same locality. In addition, better streamlining is required so that only those individuals requiring specialist opinion are referred. Taken together, this may help avoid delays in care, interruptions of patient flow and suboptimal use of resources.25 46

Several studies found pathway steps that reduced waiting times from referral to diagnosis and the number of surgeries.16 28 34 39 Unfortunately, they did not evaluate the cost, sustainability or resource implications of these steps. No studies evaluated the demand for the care pathway and the capacity available or needed to meet this demand and achieve target waiting times. Thus, no conclusions can be drawn about the potential cost–benefit or effectiveness of care, regardless of the pathway.

The majority of the available evidence is based on outcomes around professional competency for clinics containing a mixture of professional groups. A better understanding of the roles, learning, communication and expertise within this clinic structure is required, along with a robust assessment of pathway effectiveness. Although physiotherapists have been found to have high diagnostic accuracy, the complexities and diversity of the pathways presented means that future decisions on pathway design are unlikely to be as simple as replacing a surgeon with a physiotherapist.

Micro level

Measuring patient experience and using this to inform care is a key component of healthcare quality.47 Only two papers were identified that considered patient satisfaction34 35 and methodological flaws make it difficult to draw meaningful conclusions. There is limited research evidence on patient expectation and experience of hip and knee pain pathways as only one qualitative study was identified.32 The study recommended the development of a pathway with better information provision. This is an important consideration given the potential complexities and variations in care pathways that patients may experience and needs to be addressed in future studies. None of the studies evaluated differences in experience according to the location of the specialist clinic. This is a key area for patient involvement for future pathway design research.

There are some limitations of the review. First, our search for qualitative studies should also have included patient needs regarding the pathway. Second, our interpretation of clinical and cost effectiveness has been limited by a lack of randomised control trials in this area.