Outpatient physiotherapy versus home-based rehabilitation for patients at risk of poor outcomes after knee arthroplasty: CORKA RCT

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 12/196/08. The contractual start date was in August 2014. The draft report began editorial review in January 2020 and was accepted for publication in June 2020. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. This report has been published following a shortened production process and, therefore, did not undergo the usual number of proof stages and opportunities for correction. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Disclaimer

This report contains transcripts of interviews conducted in the course of the research and contains language that may offend some readers.

Copyright statement

© Queen’s Printer and Controller of HMSO 2020. This work was produced by Barker et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

2020 Queen’s Printer and Controller of HMSO

Background to the problem

Knee osteoarthritis is a common musculoskeletal condition that causes pain and loss of function. It is the most common cause of disability in older people,2 with painful osteoarthritis affecting 18.2% of people over the age of 45 years in the UK. 3 The use of KA for end-stage knee osteoarthritis is an established and effective treatment for patients who have already completed all non-surgical options but continue to experience significant pain and decreased function.

The number of KA operations taking place in the UK is continuing to rise. The 15th Annual National Joint Registry Report (2018)4 showed that over 100,000 KA operations were performed in the UK in 2017. This is an increase of over 3% from 2010. Part of the reason for this increase is that KA is increasingly performed for patients who are older and who have other health conditions in addition to their osteoarthritis. A further change relates to the increasing use of unicompartmental knee arthroplasty (UKA). Approximately 90% of KAs are a total knee arthroplasty (TKA) and 10% are a UKA or partial KA. In TKA, the entire tibiofemoral joint is replaced, whereas in UKA the affected medial or lateral side of the tibiofemoral joint is replaced only. National outcome data have revealed that patients’ age and American Society of Anesthesiologists (ASA) classification grades are increasing. In the 15th National Joint Registry Report of 2018,4 15% of patients were aged 80–89 years and 1% were over 90 years of age. The average ASA classification grade is also increasing, with 19% reported as ASA classification grade 3 (severe systemic disease that limits activity). 4

Age should not be a barrier to a good KA outcome, with reports of successful outcomes in patients aged over 80 years. 5,6

Data from patient-reported outcome measures (PROMs) have shown that most patients achieve a satisfactory outcome following KA. However, between 15% and 30% of patients are unsatisfied or report little or no improvement after KA. 7 We do not yet know how to identify these patients or target rehabilitation to improve their outcomes.

Predicting which patients will have poor outcomes

The existing literature has demonstrated that it is difficult to predict who will do well after KA. Any prediction is complex and extends far beyond a simple linear relationship with factors such as age or presurgical function. A number of studies have explored the influence of pre-operative predictors on post-operative KA outcomes and have generally agreed that patients who have a higher pre-operative status, such as better pre-operative function or less pain, tend to have better post-operative outcomes. 8–12 However, no screening tool that can accurately identify and predict who is at risk of a poor post-operative outcome currently exists, despite much previous work exploring this subject. 13–15

Rehabilitation pathway

Many hospitals follow an accelerated or enhanced recovery protocol, mobilising patients for the first time in the first 4 hours after surgery. Bohl et al. 16 and Okamoto et al. 17 demonstrated that early mobilisation within the first 24 hours after surgery shortened length of stay, but did not affect longer-term outcomes. Most hospitals have discharge criteria governing when patients are discharged from hospital that are based on safe mobilisation, ability to perform self-care and attainment of a minimum amount of knee flexion (commonly set at 90 degrees). Patients will have reached a minimal level of function at discharge, but will continue to have rehabilitation needs to help restore muscle strength and endurance, range of motion, walking distance and performance of higher-level functional activities.

Patients are commonly referred for further physiotherapy in an outpatient setting to assist with these issues. The timing of this referral varies, with some hospitals advocating immediate commencement of therapy after discharge from hospital and others waiting until the first post-operative review at around 6 weeks before considering referral. The type of therapy also differs, with some sites using self-directed home exercise and others advocating for more formal outpatient treatment using manual therapy, individually progressed exercises and functional activities in an individual or group setting.

Outpatient rehabilitation approaches

We conducted a systematic review evaluating the effectiveness of exercise after KA. We found that functional physiotherapy exercise interventions following discharge after elective primary TKA offered short-term benefit. 18 We found a small to moderate standardised effect size [0.33, 95% confidence interval (CI) 0.08 to 0.58] in favour of functional exercise at 3–4 months post operatively, which included small to moderate weighted mean differences of 2.90° (95% CI 0.61° to 5.20°) for range of motion and 1.70 (95% CI –1.00 to 4.30) for quality of life, in favour of functional exercise. However, the post-treatment benefits faded and were not carried through to 1 year post operation.

The review revealed the complexity that is involved in deciding the best rehabilitation after arthroplasty. A growing number of studies have chosen to apply an intervention later in the rehabilitation pathway (often between 6 weeks and 6 months after surgery), which suggests that a delay in starting the intervention avoids the period of early post-operative pain, swelling and limited motion. 19–21 Similarly, assessing outcomes after arthroplasty is known to be complex and multifaceted. 22,23 The lack of certainty about what constitutes best physiotherapy practice has been recognised internationally, with considerable variation in the post-operative rehabilitation protocols used. A recent Australian survey of physiotherapy practice24 and an observational cohort study of US rehabilitation services25 have started to address this knowledge gap. However, there are no published randomised controlled trials (RCTs) of occupational therapy after KA. Other studies do not allow the occupational therapy input to be disaggregated from the overarching rehabilitation package to assess its contribution.

Patients in the UK usually receive a short course of between four and six sessions of post-operative physiotherapy after their surgery, usually in a physiotherapy outpatient clinic setting. Concern has been raised that many exercise programmes lack adequate intensity to lead to optimal recovery. 24,26 Internationally, where much longer courses of physiotherapy are often provided, research has indicated that 12–18 hours of physiotherapy21 or a mean of 17 visits27 may be needed to produce benefit. These levels of care are substantially higher than those provided in the UK and, in the current economic climate, may be more than the NHS can afford given the rising number of KA operations that are performed each year. Patients are likely to have an occupational therapy assessment before surgery to identify any potential issues on discharge, such as home layout. However, it is not usual for patients to have further input from occupational therapists unless they have particular problems.

Given the rising number of KA operations, the relatively limited therapy resources available and the increasing age and frailty of patients receiving this surgery, it is important that we concentrate our rehabilitation resources on those patients who need the most help to achieve a good outcome. It is important that exercise and functional rehabilitation is linked to demonstrable increases in function and participation levels. Not all patients need physiotherapy to help them recover, and some patients will recover fully using self-directed rehabilitation. 28–31 It is, thus, pertinent to focus our resources on those who are most likely to be at risk of a poor outcome, least likely to be able to engage with a self-management approach and able to benefit from rehabilitation input.

It is clear that current rehabilitation strategies do not meet the needs of all patients, particularly those who are socially isolated, do not have easy access to transport and are frail. This is a particular concern given both the projected increased need for joint arthroplasty over the next decade to accommodate an ageing population and the pressure of potential reductions in NHS funding. Evaluating the value of treatment modalities offered to these patients is crucial because many more patients are being discharged home earlier from the acute setting. There is less time available for acute physical recovery, rehabilitation and education in hospital, which increases the potential burden of care for these patients and their families.

Rationale for the CORKA trial

There is considerable variation in the quantity and content of rehabilitation received by patients after KA. Previous evidence has shown that some patients do well with minimal input or self-directed exercise programmes,30–33 but that many patients continue to experience pain and poor function after their surgery and report dissatisfaction with the rehabilitation that they have received.

In the CORKA trial, we aimed to evaluate a different approach designed to cater to the needs of an older, frailer population. We evaluated two strategies of rehabilitation after KA: a traditional outpatient clinic-based model and a rehabilitation programme delivered in participants’ homes by rehabilitation assistants supported by qualified therapists.

The programmes were designed to be acceptable to NHS physiotherapists and occupational therapists, based on current available evidence, and affordable to UK commissioners. A parallel economic study enabled conclusions to be made about cost-effectiveness. A nested qualitative study assessed the acceptability of the interventions to participants and the therapists who delivered them.

Research objectives

The CORKA research objectives were published in the protocol paper in Trials in 2016:34

1. to design a prognostic screening tool based on an analysis of factors associated with poor outcomes following KA to guide patient selection for the trial

2. to evaluate if a multicomponent rehabilitation programme delivered in patients’ homes could improve their outcomes compared with those receiving standard outpatient rehabilitation over 12 months

3. to undertake a nested qualitative study exploring patients’ and clinicians’ perceptions of the community-based rehabilitation programme

4. to undertake an economic analysis to compare the cost-effectiveness of the interventions.

Trial design

The CORKA trial was a prospective, individually randomised, two-arm controlled trial with blinded outcome assessment for the clinical outcomes at baseline, 6 months and 12 months. It aimed to determine if a multicomponent rehabilitation programme provided to participants who were deemed at risk of a poor outcome following KA by the CORKA screening tool (see Chapter 7) was better than usual care. The trial took place in 14 NHS hospitals across the UK. The design included a parallel health economic evaluation and a nested qualitative study. Figure 1 gives an overview of the flow of participants through the trial and the timing of assessment procedures.

FIGURE 1.

Flow of participants through the trial.

Participants

Interventions

Full details of the interventions are provided in Chapter 3, but are described briefly here. This trial tested physiotherapy and occupational therapy in a multicomponent package of rehabilitation that was delivered by a generic rehabilitation therapist. The occupational therapy element focused on assessment of and adaptations to participants’ homes to enable a safe environment for home exercise and everyday functional tasks. The physiotherapy element included a home-based intervention with an emphasis on functional, activity-based rehabilitation. We tested whether or not the intervention could be delivered by generic rehabilitation assistants rather than uni-professionally.

Usual care can vary geographically, including the number of sessions of physiotherapy given post discharge. 36 To standardise the usual-care arm without changing it significantly, we set a minimum and maximum number of sessions allowed for usual care. Participants were expected to attend at least one and a maximum of six sessions of outpatient physiotherapy.

Data collection

Outcomes

Follow-up data collection was carried out at face-to-face clinical assessments at 6 and 12 months following randomisation. Where face-to-face assessment was not possible, postal and telephone data collection methods were used to obtain self-reported core data. Table 1 shows the measures collected at each time point.

Intervention monitoring and support

Quality assurance checks took place at all CORKA research sites. They included fidelity checks, during which assessment and treatment delivery were observed. All aspects of the intervention were checked against a predefined fidelity checklist that was created in line with the study protocol. Following checks, feedback was given and any training needs or action points identified. The investigator site file was also reviewed and any necessary changes were recorded. Staff at research sites were assured that support could be provided as required, not just during a monitoring visit. Staff were encouraged to contact the central CORKA trial team in Oxford if they had any questions or concerns about any aspect of the trial.

Randomisation

Randomisation took place using the Registration/Randomisation and Management of Product website provided by the Oxford Clinical Trials Research Unit randomisation service.

Sample size

We chose the LLFDI as the primary outcome using the overall function subscore at 1 year. The sample size calculation was based on a moderately small standardised effect size of 0.275. This standardised effect size is, for example, equivalent to detecting a three-point difference between treatment arms on the LLFDI overall function score, assuming a standard deviation (SD) of 10.91 and no clustering effect across sites. 45

A total of 558 participants (279 per arm) would be required to detect a standardised effect size of 0.275, with 90% power and 5% (two-sided) significance. A 10% loss to follow-up was chosen based on previous experience of trials in a similar population, which resulted in the planned sample size of 620 participants (310 per arm).

An internal pilot study was conducted at one site (Oxford) to review recruitment feasibility and confirm the intervention package. Fifteen participants were randomised during this pilot study. These participants are included in the final analysis, as the interventions were not changed between the pilot and the main trial.

Data analysis

Monitoring and approvals

Overview of the development process

The CORKA home-based intervention development drew on a number of sources, including reviewing the relevant evidence base on post-arthroplasty rehabilitation, identifying appropriate exercise guidelines and seeking the views of relevant stakeholders. An intervention development day was organised that was attended by research staff, physiotherapists and occupational therapists. During the day, potential intervention approaches were discussed and refined, and factors associated with the design, content and intervention delivery were discussed. A draft intervention was presented.

The draft intervention was reviewed and refined, then tested in a clinical NHS setting as part of a pilot phase, which included participant and clinician feedback.

The development steps are outlined below and are represented in Figure 2.

FIGURE 2.

Intervention development considerations.

Health Technology Assessment programme brief

The first step in developing the CORKA home-based intervention was considering the HTA commissioning brief. This brief called for an intensive and, if possible, multidisciplinary rehabilitation intervention in the community setting that was designed for older adults who were at risk of poorer functional outcomes following elective knee joint arthroplasty surgery that could be compared with usual care.

Evidence base

Before designing a draft intervention, we undertook a systematic search of the literature on rehabilitation post-KA, and a wider search on the field of rehabilitation following arthroplasty. This process identified a number of papers that were used in the design of the intervention. These papers are outlined in Table 2 and are described in more detail in the following section on the rationale underlying the CORKA home-based intervention.

Rationale underlying the CORKA intervention

The different components of the CORKA home-based intervention were informed by a review of existing trials. The intervention primarily consists of a functional exercise programme. The rationale for the following sections of this exercise programme is given below:

• range of movement exercises

• strengthening exercises

• balance exercises

• gait and aerobic exercise

• functional exercise.

Range of movement

Both flexion and extension can be limited after KA surgery. 58,79 Stiffness is a common complication after surgery,78 with as many as 60% of those undergoing KA experiencing stiffness. 73 Many functional activities require a knee joint range of 0–110°,80 and a significant proportion of this range is likely to be attained within the first 12 weeks after surgery. 79 A number of studies have, therefore, included range of movement exercises in their post-arthroplasty interventions. 56,66,75

Strengthening

Decreased quadricep strength is a common finding in people with knee osteoarthritis,80 and these strength deficits can persist for a long time after KA. 76 One year after arthroplasty, both knee flexor and extensor peak muscle torque are lower in those who have undergone surgery than age-matched controls. 83 As there is a close link between muscle strength and functional performance, strength is an important consideration for the population in question. For example, the ability to change position from sitting to standing and the ability climb stairs have been linked to quadriceps strength. 70 The ability to get up from a chair, climb-up stairs, walk and change direction have been linked with hip abduction strength. 75 Hip and calf strength also perform a significant role in gait and balance. 65,73 Strengthening exercises have formed part of previous interventions following lower limb arthroplasty, targeting muscle groups such as the quadriceps, hamstrings, hip abductors and calf muscles. 21,56,59,75

Balance

As KA surgery can affect gait, balance and stability,67 it is not surprising that falling has been reported as a problem after KA. 63 In a pilot study, Piva et al. 19 gave participants functional training or functional training enhanced by a balance training programme after KA. The pilot study’s design was underpowered to detect changes between groups but showed that the balance training group had high adherence, had a low dropout rate and reported no adverse events. A RCT66 that investigated the effect of adding a balance programme to functional training after KA reported that those in the balance training group had significantly better outcomes, including on the Timed Up and Go, 10-metre walk, 30SCST and Stair Climb tests.

Gait skills and aerobic exercise

McClelland et al. 68 conducted a systematic review that studied gait analysis following KA and concluded that patients exhibit altered gait patterns after surgery. Gait problems after KA include specific gait deficits linked to underlying problems, such as reduced range of movement or pain, and general problems, such as decreased walking speed and endurance. 62,68,83 Such gait problems may continue to persist 1 year after surgery. Patients undergoing UKA are more likely to exhibit a gait pattern closer to normal than patients undergoing TKA. 84 It is important to consider gait in patients who have undergone KA, as gait outcomes influence patient satisfaction. 82

Bruun-Olsen et al. 20 undertook a RCT comparing a walking skills intervention with usual physiotherapy. They reported that the walking skills group demonstrated better short-term and long-term functional mobility. Walking can also be considered from a wider cardiovascular perspective. People with end-stage lower limb osteoarthritis are likely to be physically deconditioned,74,77 which can be exacerbated immediately after surgery owing to reduced mobility. Naylor and Ko71 found that moderate-intensity exercise was tolerable and safe following KA. Later studies have included aerobic exercise using cycling on a static bike or walking on a treadmill. 59,61

Functional exercise

Functional limitations can persist after the immediate post-operative period following KA. 57,69 One year post KA surgery, it is common for patients to be significantly slower with activities, such as walking or stair climbing, and have more functional impairments with tasks that involve kneeling or squatting than matched controls. 72,83 Measurements of function, such as the Timed Up and Go, 6-minute walk, Stair Climb and SLS tests, have been found to be worse in participants following KA at 1, 3 and 6 months after surgery than in healthy participants. 55 A 2007 systematic review18 investigating the effectiveness of post-KA exercise advocated for the use of functional physiotherapy exercise interventions to achieve short-term benefits. It found that the interventions had a small to moderate effect on functional outcomes, QoL and joint range of movement at 3 to 4 months after surgery; however, these benefits were not found 1 year after surgery. A later study explored function after KA57 and a 2013 systematic review76 of exercise after KA recommended that physiotherapy interventions include strengthening and functional exercise.

Guidelines

Currently, there are no specific exercise guidelines for patients after KA. We considered two relevant guidelines for patients following KA when developing the CORKA home-based intervention:

• Physical activity and public health in older adults: recommendation from the American College of Sports Medicine and the American Heart Association. 85

• Quantity and quality of exercise for developing and maintaining cardiorespiratory, musculoskeletal, and neuromotor fitness in apparently healthy adults: guidance for prescribing exercise. 86

These guidelines cover exercise guidelines for adults aged 65 years or older and aged 50–64 years with functional limitations or a chronic condition,85 and apparently healthy adults of all ages. 86 On the basis of these guidelines, it is recommended that participants accumulate five sessions of 30 minutes of moderate-intensity physical activity per week, which is accumulated in at least 10-minute bouts.

Intervention development day and workshop

An intervention development day was attended by researchers, physiotherapists and occupational therapists. A draft intervention developed using the evidence base and guidelines outlined above was presented at the workshop. The proposed intervention programme, strategies for exercise progression and patient materials were discussed and reviewed. Feedback was given on all aspects of the proposed intervention, leading to changes in the content of the intervention and patient materials.

Pilot phase

Following the workshop, a final version of the CORKA home-based intervention was compiled. It was then tested in a pilot phase with 15 participants. Three participants and two clinicians were interviewed after the participants had completed their treatment sessions to get feedback on the intervention content and the mode of delivery. As a result, changes were made to the participant materials and intervention procedures. These modifications included retaking some of the photographs used in the participant exercise sheets to make them brighter and more vibrant, and providing a space on the sheets for participants to make comments.

The CORKA intervention arms

The CORKA trial had two intervention arms:

• usual care

• the CORKA home-based intervention.

The intervention arms are described in the following sections.

The CORKA home-based intervention

The CORKA home-based intervention was an individually tailored, multicomponent rehabilitation programme that could be adapted for each participant. The intervention’s aim was to improve the function and participation of participants who were at risk of a poor outcome post KA. The primary component was an individually adapted exercise programme that was conducted in the participant’s own home, with additional components consisting of functional task practice, appropriate adherence approaches and, if required, the provision of appropriate aids and equipment.

The CORKA home-based intervention started within 4 weeks of surgery. It comprised an initial home assessment appointment, followed by up to six follow-up sessions. It was designed to be delivered by a mixture of qualified occupational therapy and physiotherapy staff, and rehabilitation assistants. The CORKA home-based intervention is outlined further in Figure 3.

FIGURE 3.

Components of the CORKA home-based intervention exercise programme. ROM, range of movement.

Home exercise programme

The CORKA home-based intervention exercise programme comprised groups of possible exercises that were arranged into the following categories:

• knee flexion range of movement

• knee extension range of movement

• basic quadriceps strengthening

• strengthening – quadriceps

• strengthening – hamstrings

• strengthening – hip abductors

• strengthening – calf

• balance

• gait skills.

The programme included a range of exercises in each category, and at least one exercise from each category was to be selected for a participant’s exercise programme. This allowed the intervention to cover a range of exercises and be tailored to each participant.

Safety and serious adverse events

The safety of participants and therapists was paramount and was considered as part of the CORKA home-based intervention. All trial site staff members delivering the intervention were encouraged to consider their and their patients’ safety. One of the significant considerations for a community-based trial like CORKA is that staff are by themselves when visiting participants in their homes. All staff members at all sites were, thus, asked to adhere to their organisation’s lone-working policy. They were also asked to report anything that they felt might constitute a serious adverse event. To make this process as easy as possible for clinicians involved in the treatment sessions, they were encouraged to discuss any questions or concerns with the central CORKA team in Oxford.

Study participants

The trial opened for recruitment on 17 March 2015, and the first participant was randomised on 27 March 2015. As recruitment was slower than predicted in the first year, the funder (HTA programme) requested a recovery plan be put in place. The recovery plan was reviewed by the Trial Steering Committee on 16 June 2016 and submitted to the HTA programme on 8 July 2016. In February 2017, a funding extension was agreed with the HTA programme with a revised trial end date of 31 December 2019.

The final randomisation took place on 26 January 2018. In total, 2788 patients were screened and 621 participants were recruited to the trial. The 12-month follow-up was completed in February 2019. The CONSORT flow chart (Figure 4) summarises the flow of participants through the trial, including details of the number of participants randomised and the numbers allocated to and receiving at least one session of each treatment. The number of participants followed up at 6 and 12 months and providing the primary outcome measure is summarised by treatment group. Reasons why participants were not followed up are also included. The numbers of participants included in the ITT and PP populations are summarised by treatment group.

FIGURE 4.

The CONSORT flow chart. LTFU, lost to follow-up. a, Clinical withdrawals include those who withdrew owing to an adverse event, serious adverse event or medical contraindication; b, missing follow-up data in the ITT population were imputed using multiple imputation for the primary outcome analysis; c, analysis of the PP population was based on available data only.

Baseline characteristics

Randomisation in the CORKA trial was stratified by recruiting site. The number and proportion of participants randomised to each treatment group at each site are summarised in Table 6. Participants were well balanced between the two groups at each site.

A screening tool was used to identify participants at higher risk of a poor outcome following KA for inclusion in the CORKA trial (see Chapter 7). The key features of the screening tool were body mass index (BMI), pain, health status and anxiety or depression.

Table 7 summarises the number and proportion of people satisfying each of the criteria, which combine to give the total screening tool score. These appear well balanced across the two treatment groups. The slight apparent imbalance in BMI may have been because this variable was treated as categorical (see Table 8). Table 7 also included the contribution of each characteristic to the total score, the number of participants with each score and the constituent parts that make up that score. All participants met the moderate to severe usual pain from the knee criteria, which was usually combined with a high BMI.

The descriptive characteristics of all randomised participants are summarised by treatment group, and overall, in Table 8. These values are presented as numbers and percentages for binary and categorical factors, and as means and SDs or medians and IQRs, as appropriate, for continuous variables. These variables appear well balanced across the two treatment groups.

The PROMs and physical measure baseline values are summarised by treatment group, and overall, in Table 9, and are all similar across the two treatment groups.

Treatment compliance

Participants were defined as complying with usual-care if they attended at least one treatment session. Compliance with home-based rehabilitation was defined as receiving at least four treatment sessions. Table 10 summarises the number and percentage of compliers in each treatment group, the number of participants who attended no sessions and the average number of treatment sessions received.

Compliance was high in both groups. The median number of treatment sessions was higher in the home-based rehabilitation arm than in the usual-care arm. However, there was significant variability in the number of sessions of usual care received, ranging from 0 to 27 sessions. Participants who underwent UKA and TKA attended similar average (median, IQR) numbers of sessions. Figure 5 shows how many participants in each treatment group received each number of sessions.

FIGURE 5.

Comparison of number of treatment sessions received in the two treatment groups.

The time from randomisation to first treatment session is also summarised in Table 10. Most participants received their first session within 4 weeks, as per the protocol. Similar numbers of participants received their first treatment within 4 weeks in the two treatment arms.

The average (median and IQR) and minimum and maximum number of treatment sessions received were calculated separately by treatment group for each recruiting site (Table 11). The ‘other’ category for recruiting sites includes all sites that recruited < 15 participants. Sites 4 and 5 provided follow-up care jointly and are, therefore, combined in this table.

Substantial variability by site in the number of treatment sessions received was identified. The median number of sessions of usual-care treatment ranged from just two sessions at site 13 to six sessions at site 1. Similarly, the median number of sessions of home-based rehabilitation varied from four sessions at site 1 to seven sessions at sites 4 and 5 and site 10. Site 1 participants who were allocated to usual care received more treatment sessions, on average, than those allocated to home-based rehabilitation.

Details of the usual care delivered by each site are summarised in Table 12. Again, the ‘other’ sites are grouped together and sites 4 and 5 are combined. The total number of sessions provided, the number of participants recruited and the average number of sessions are given for each site. The number and proportion of sessions taking place in each location (home and hospital) are summarised by site. Although > 80% of sessions took place in hospital at most sites, only 21.6% of sessions took place in hospital at site 13, with the rest of the usual-care patients seen in a community setting.

The number and proportion of sessions of individual therapy provided at each site are summarised. The sites can be separated into two groups. At sites 1 and 11 and the ‘other’ sites, the sessions were split relatively evenly between individual therapy (40% to 60% of participants) and group therapy. At the remaining sites, > 80% of the sessions were individual therapy. Most sessions at all sites were exercise therapy (> 85%).

Details of the home-based rehabilitation treatment delivered by each site are summarised in Table 13. The number of participants and the total number of treatment sessions given are included in this table. The number and proportion of treatment sessions, including each type of exercise, appear similar across sites, with some notable exceptions. A smaller proportion of sessions delivered by site 1 than the rest of the sites included flexion, extension and basic quadriceps exercises, and a smaller proportion of sessions delivered by site 13 than the rest of the sites included static balance and gait skills exercises.

Primary outcome analyses

Trends in LLFDI function scores over time are plotted for each treatment group in Figure 6. There was a substantial improvement between baseline and 6 months in both groups, with minimal additional difference between 6 and 12 months post randomisation. No substantial differences between the two groups are visible at either time point.

FIGURE 6.

The LLFDI function score from baseline to 12 months post randomisation for each treatment group.

The primary analysis of the LLFDI function score was carried out for the ITT population using multiple imputation. There was no statistically significant difference between the two treatment groups at the primary time point of 12 months post randomisation (adjusted difference 0.49, 95% CI –0.89 to 1.88; p = 0.34) or at 6 months (Table 14).

Sensitivity analyses were carried out considering the ITT population using available cases only, the PP population using available cases and a complier-average causal effects analysis. No significant differences between the two treatment groups at either time point were identified (see Table 14). Additional sensitivity analyses of the primary outcome included unadjusted models; models with no treatment-by-time interaction, including time as a continuous rather than categorical predictor; and models excluding data recorded outside the specified time windows. None led to a different conclusion.

Subgroup analyses

Consistency of the treatment effect across sites was explored by including an interaction between treatments and recruiting site in the model. A forest plot of treatment effects for each recruitment site is provided in Figure 7. The treatment effects varied across sites, from an effect of 9 points in favour of home-based rehabilitation at site 10 (95% CI 2.95 to 15.10 points) to 1 point in favour of usual care at site 1 (95% CI –4.80 to 2.86 points). This variability appears to be at least in part related to the amount of treatment received at each site, with participants in site 1 receiving more sessions of usual care and less sessions of home-based rehabilitation than those in site 10.

FIGURE 7.

Forest plot of treatment effects by recruiting site. Sites 4 and 5 provided follow-up jointly, so are reported together. ‘Other’ sites include all those that recruited fewer than 15 participants.

Similar analyses were used to explore the relationship between treatment effect and age, gender, number of falls, screening tool score and the constituent parts of the screening tool (BMI category, health status and anxiety/depression). No clear patterns were identified. A forest plot of treatment effects by screening tool score is provided in Appendix 1.

Secondary outcome analyses

Patient-reported outcome measures

Trends over time for each of the key secondary PROMs are plotted in Figure 8. Although there was only a small improvement over time in the two LLFDI disability variables, the OKS and KOOS both showed similar substantial improvement between baseline and 6 months in both treatment groups. There was little additional improvement between 6 and 12 months for either variable.

FIGURE 8.

Key secondary participant-reported outcome measures from baseline to 12 months post randomisation by treatment group. (a) LLFDI disability frequency; (b) LLFDI disability limitation; (c) OKS; and (d) KOOS.

Trends over time are plotted for the PASE and EQ-5D-5L secondary outcomes in Figure 9, showing small improvements over time.

FIGURE 9.

Additional secondary participant-reported outcome measures from baseline to 12 months post randomisation by treatment group. (a) PASE; (b) EQ-5D-5L utility; and (c) EQ-5D-5L visual analogue scale.

The primary analysis for each of the secondary PROMs was conducted for the ITT population using available cases only. No significant differences between the two treatment groups at either time point (6 and 12 months) were identified for any of the secondary outcomes (Table 16).

TABLE 16 - Comparison of treatment groups for secondary outcomes (intention-to-treat population using available cases)

VAS, visual analogue scale.

Notes

Each LLFDI disability scale ranges from 0 to 100, with higher scores indicating less disability. OKS scores range from 0 to 48, with lower scores indicating better knee function. KOOS ranges from 0 to 100, with higher scores indicating better QoL. For PASE, higher scores indicate greater levels of physical activity. EQ-5D-5L utility scores range from –0.594 to 1, and the EQ-5D-5L visual analogue scale ranges from 0 to 100, in both cases higher scores indicate better QoL.

Time point	Usual care		Home-based rehabilitation		Adjusted difference (95% CI)	p-value
	Mean (SD)	Total	Mean (SD)	Total
LLFDI disability (frequency)
Baseline	51.28 (7.25)	311	51.55 (7.46)	308	–	–
6 months	54.27 (6.86)	283	54.91 (6.85)	275	0.64 (–0.50 to 1.78)	0.27
12 months	54.13 (6.87)	287	55.05 (6.87)	279	0.93 (–0.21 to 2.06)	0.11
LLFDI disability (limitation)
Baseline	66.26 (11.96)	312	66.25 (12.48)	305	–	–
6 months	72.33 (15.25)	284	74.72 (15.24)	275	2.39 (–0.14 to 4.92)	0.06
12 months	74.10 (15.27)	287	74.75 (15.28)	277	0.65 (–1.87 to 3.18)	0.61
OKS
Baseline	20.59 (7.50)	312	20.81 (7.31)	308	–	–
6 months	35.69 (7.69)	284	36.23 (7.69)	277	0.54 (–0.73 to 1.81)	0.40
12 months	37.34 (7.71)	287	37.80 (7.71)	278	0.46 (–0.81 to 1.73)	0.48
KOOS – QoL subscale
Baseline	25.50 (16.58)	301	25.50 (15.89)	300	–	–
6 months	61.26 (22.31)	273	62.47 (22.29)	269	1.21 (–2.55 to 4.97)	0.53
12 months	65.53 (22.37)	277	65.26 (22.36)	271	–0.27 (–4.02 to 3.47)	0.89
PASE
Baseline	121.24 (73.90)	302	115.11 (72.95)	297	–	–
6 months	150.62 (66.72)	259	149.41 (66.46)	245	–1.22 (–12.86 to 10.42)	0.84
12 months	156.58 (66.82)	263	154.63 (66.73)	251	–1.95 (–13.51 to 9.61)	0.74
EQ-5D-5L utility
Baseline	0.52 (0.22)	308	0.52 (0.23)	308	–	–
6 months	0.75 (0.18)	277	0.75 (0.18)	275	0.00 (–0.03 to 0.03)	0.85
12 months	0.76 (0.18)	282	0.76 (0.18)	278	–0.00 (–0.03 to 0.03)	0.99
EQ-5D–5L VAS
Baseline	68.53 (19.46)	312	68.03 (19.13)	308	–	–
6 months	75.81 (14.55)	283	75.51 (14.54)	275	–0.30 (–2.71 to 2.11)	0.81
12 months	76.80 (14.57)	287	77.05 (14.58)	278	0.24 (–2.16 to 2.65)	0.84

For the key secondary PROMs (LLFDI disability frequency and limitation, OKS and KOOS QoL), sensitivity analyses were performed for the PP population using available cases and for the ITT population using multiple imputation, and complier-average causal effects were calculated. Only one significant difference was identified, which was for the LLFDI disability limitation at 6 months using the ITT population and multiple imputation (Table 17). This effect only just reached significance (adjusted difference 2.67, 95% CI 0.14 to 5.19; p = 0.04) and is probably a chance effect.

TABLE 17 - Sensitivity analyses of key secondary participant-reported outcome measures

Time point	Usual care		Home-based rehabilitation		Adjusted difference (95% CI)	p-value
	Mean (SD)	Total	Mean (SD)	Total
Per-protocol population using available cases
LLFDI disability (frequency)
Baseline	51.22 (7.41)	243	51.74 (7.50)	293	–	–
6 months	54.39 (6.96)	223	55.03 (6.96)	269	0.63 (–0.60 to 1.87)	0.32
12 months	54.37 (6.97)	229	55.09 (6.97)	270	0.72 (–0.51 to 1.95)	0.25
LLFDI disability (limitation)
Baseline	65.87 (11.25)	244	66.30 (12.39)	291	–	–
6 months	73.12 (15.35)	224	74.73 (15.37)	269	1.60 (–1.13 to 4.33)	0.25
12 months	74.44 (15.38)	229	74.99 (15.38)	268	0.55 (–2.17 to 3.27)	0.69
OKS
Baseline	20.57 (7.41)	244	20.98 (7.32)	293	–	–
6 months	36.23 (7.53)	224	36.34 (7.56)	271	0.11 (–1.22 to 1.45)	0.87
12 months	37.96 (7.57)	229	37.82 (7.56)	269	–0.14 (–1.48 to 1.19)	0.83
KOOS – QoL subscale
Baseline	25.53 (16.67)	235	25.66 (15.83)	285	–	–
6 months	63.14 (22.00)	215	62.70 (22.05)	263	–0.44 (–4.41 to 3.54)	0.83
12 months	67.40 (22.08)	220	65.76 (22.06)	262	–1.64 (–5.60 to 2.32)	0.42
Intent-to-treat population using multiple imputation
LLFDI disability (frequency)
Baseline	51.28 (7.24)	312	51.52 (7.46)	309	–	–
6 months	53.954 (7.284)	312	54.58 (8.09)	309	0.63 (–0.54 to 1.79)	0.29
12 months	53.81 (7.13)	312	54.63 (7.73)	309	0.82 (–0.32 to 1.96)	0.16
LLFDI disability (limitation)
Baseline	66.26 (11.96)	312	66.19 (12.49)	309	–	–
6 months	71.97 (15.88)	312	74.51 (15.62)	309	2.67 (0.14 to 5.19)	0.04
12 months	73.80 (16.63)	312	74.51 (15.62)	309	0.71 (–1.88 to 3.30)	0.59
OKS
Baseline	20.59 (7.50)	312	20.82 (7.30)	309	–	–
6 months	35.51 (8.69)	312	36.05 (8.42)	309	0.54 (–0.80 to 1.89)	0.43
12 months	37.18 (8.50)	312	37.80 (8.04)	309	0.62 (–0.70 to 1.94)	0.35
KOOS – QoL subscale
Baseline	25.49 (16.33)	312	25.66 (15.70)	309	–	–
6 months	61.09 (23.51)	312	62.50 (24.60)	309	1.41 (–2.36 to 5.18)	0.46
12 months	65.22 (23.64)	312	65.24 (23.59)	309	0.02 (–3.62 to 3.66)	0.99
Complier-average causal effect
LLFDI disability (frequency)
Overall	–	–	–	–	0.87 (–0.22 to 1.97)	0.12
LLFDI disability (limitation)
Overall	–	–	–	–	1.69 (–0.79 to 4.17)	0.18
OKS
Overall	–	–	–	–	0.56 (–0.76 to 1.88)	0.41
KOOS – QoL subscale
Overall	–	–	–	–	0.51 (–3.28 to 4.31)	0.79

Physical measures

Trends over time in the physical measures are summarised in Figure 10. There were small improvements over time in each measure, but none was substantial.

FIGURE 10.

Physical measures from baseline to 12-month follow-up by treatment group. (a) 30SCST number of stands; (b) F8WT time in seconds; (c) F8WT number of steps; (d) arthroplasty side stance time; and (e) contralateral side stance time.

The physical measures are summarised by treatment group over time in Table 18. No statistically significant differences between the two groups were identified in the 30SCST or F8WT. There was a statistically significant difference in the SLS test between the stance time at baseline and the stance time at 12 months on the contralateral leg (p = 0.03). However, it is difficult to see how this change could be related to the treatment received, and is likely to have been because of improved overall fitness and strength.

TABLE 18 - Comparison of physical measures between treatment groups at follow-up time points

The 30SCST records the number of stands in 30 seconds, with higher scores indicating better function. For the F8WT a quicker time and fewer steps indicate better function. In the SLS, stance time is averaged over three trials at standing on the same leg, with a maximum score of 45 seconds and higher scores indicate better function.

Time point	Usual care (N = 312)		Home-based rehabilitation (N = 309)		Adjusted difference (95% CI)	p-value
	Summary	n	Summary	n
30SCST number of stands, mean (SD)
Baseline	8.2 (3.5)	312	8.4 (3.7)	309	–	–
6 months	11.1 (3.1)	278	11.0 (3.1)	275	–0.06 (–0.58 to 0.46)	0.82
12 months	11.7 (3.1)	279	11.5 (3.1)	276	–0.22 (–0.74 to 0.3)	0.41
F8WT time (seconds), mean (SD)
Baseline	11.8 (5.4)	312	12.1 (4.9)	309	–	–
6 months	10.2 (3.6)	277	9.6 (3.6)	275	–0.57 (–1.17 to 0.04)	0.07
12 months	9.3 (3.6)	280	9.1 (3.6)	274	–0.17 (–0.78 to 0.43)	0.58
F8WT steps, mean (SD)
Baseline	16.4 (4.5)	312	17.3 (5.0)	309	–	–
6 months	15.5 (3.2)	277	15.2 (3.2)	275	–0.36 (–0.91 to 0.18)	0.19
12 months	15.2 (3.2)	280	14.9 (3.3)	274	–0.23 (–0.77 to 0.31)	0.40
KA leg SLS average time (seconds), median (IQR)
Baseline	5.3 (2.6, 13.0)	310	4.5 (1.9, 14.5)	308	–	–
6 months	10.0 (3.9, 30.3)	279	10.0 (3.6, 26.8)	276	–	0.57
12 months	13.3 (5.9, 31.1)	280	11.7 (3.4, 31.1)	276	–	0.14
Contralateral leg SLS average time (seconds), median (IQR)
Baseline	7.2 (3.6, 22.0)	310	6.3 (2.6, 18.0)	308	–	–
6 months	10.9 (4.2, 30.8)	279	9.8 (3.5, 29.7)	276	–	0.43
12 months	14.7 (5.6, 30.7)	280	11.0 (3.2, 29.7)	276	–	0.03

Details of adaptations used to perform the 30SCST are summarised by treatment group and follow-up time point in Table 19. Fewer participants required an adaptation as time went on. This table also summarises the proportion of participants who stayed within the cones in the F8WT (high throughout) and the F8WT smoothness scores, which improved over time.

TABLE 19 - Further details on the 30SCST and F8WT

Higher smoothness scores indicate better performance.

	Usual care (N = 312), n (%)			Home-based rehabilitation (N = 309), n (%)
	Baseline	6 months	12 months	Baseline	6 months	12 months
30SCST adaptations
None	211 (67.6)	224 (80.6)	231 (82.8)	214 (69.3)	229 (83.3)	223 (80.8)
Uses hands on legs	94 (30.1)	50 (18.0)	47 (16.8)	90 (29.1)	44 (16.0)	49 (17.8)
Uses walking aid	2 (0.6)	2 (0.7)	1 (0.4)	4 (1.3)	1 (0.4)	1 (0.4)
Not tested: unable	5 (1.6)	1 (0.4)	0 (0.0)	1 (0.3)	1 (0.4)	3 (1.1)
Not tested: refused	–	1 (0.4)	–	–	0 (0.0)	–
F8WT
Stayed within cones?	303 (97.1)	274 (98.9)	273 (97.8)	304 (97.4)	273 (98.6)	270 (96.8)
F8WT smoothness score
0	9 (2.9)	5 (1.8)	2 (0.7)	11 (3.6)	5 (1.8)	4 (1.5)
1	73 (23.4)	26 (9.4)	33 (11.8)	87 (28.2)	32 (11.6)	26 (9.5)
2	50 (16.0)	35 (12.6)	22 (7.9)	56 (18.1)	29 (10.5)	27 (9.9)
3	180 (57.7)	211 (76.2)	222 (79.6)	155 (50.2)	209 (76.0)	217 (79.2)

Additional analyses

The Trial Steering Committee and Data Safety Monitoring Committee met jointly to review the results on 13 June 2019. At this meeting they requested that additional analyses be performed that explored further the definition of the PP population. The committees queried whether or not the key part of the new intervention might be the home-based nature of the treatment, rather than the exact number of sessions. They requested that the analysis of the primary outcome be repeated using two more definitions of the PP population:

1. excluding those who received a session of usual-care treatment at home

2. excluding all participants randomised at site 13, at which the delivery of usual care differed significantly from other sites.

These additional analyses were performed and the results are presented here.

Methods

Analysis

We report descriptive statistics (means and SDs) for resource use, costs and EQ-5D-5L utilities at each follow-up time point using complete data only. Differences between arms were estimated using linear regression controlling for follow-up period, treatment allocation, an interaction between follow-up time and treatment allocation, recruitment site, and baseline utility score for EQ-5D-5L. Standard errors were adjusted for multiple observations within individuals.

Following multiple imputation, we estimated total costs and QALYs for all 621 participants in the CORKA trial from the date of trial recruitment to the earliest date of death or withdrawal from the trial, or the end of follow-up at 1 year by treatment allocation. On each imputed data set we estimated mean costs (by type) and QALYs using separate linear regression models controlling for treatment allocation, recruitment site and for QALYs baseline EQ-5D-5L utility. Estimates derived from each imputed data set were combined using Rubin’s rule to estimate the adjusted mean difference and standard error for each outcome. As a sensitivity analysis, we explored using a complete-case analysis that included only individuals who provided complete data over the 12-month trial duration. In the preliminary analysis, we also considered using linear regression in imputation models rather than predictive mean matching.

We calculated the ICER by dividing the mean cost differences (total NHS costs and total societal costs) between the CORKA home-based intervention and usual care by the mean QALY difference. We estimated the joint uncertainty around incremental total costs and QALYs (i.e. the difference between the CORKA home-based intervention and the usual-care arm) and in the cost-effectiveness by bootstrapping 200 times from each of our 40 imputed data sets (creating 8000 bootstraps), running the estimation model on each bootstrapped data set and extracting the estimated treatment effects. From these bootstrapped results, we calculated the probability that the CORKA home-based intervention was more cost-effective than usual care for different threshold values per QALY gained,104 by estimating the proportion of bootstrap replicates with a net monetary benefit of above zero for each threshold value. The net monetary benefit was given by the product of the mean difference in QALYs and the threshold value minus the mean difference in costs.

All analyses were conducted using R version 3.4 (The R Foundation for Statistical Computing, Vienna, Austria).

Results

EuroQol-5 Dimensions, five-level version, utility

Table 24 presents EQ-5D-5L utility scores and differences by treatment allocation at each time point. EQ-5D-5L scores were very similar in the two treatment groups at all follow-up time points, improving similarly between baseline and 6 months. The distribution of responses to each EQ-5D-5L domain at each follow-up time point is presented by treatment allocation in Appendix 3, Table 44.

Main analysis

Table 25 shows the main analysis results at 1 year. Appendix 3, Table 45, shows descriptive statistics for the costs and QoL at each follow-up time point after multiple imputation. There was a small, non-significant difference in QALYs (0.003, 95% CI –0.017 to 0.023) favouring the CORKA home-based intervention.

Post-operative physiotherapy (intervention) costs were lower on average in the CORKA trial arm than in usual care (–£65, 95% CI –£86 to –£44). However, as the CORKA trial arm reported higher subsequent health-care costs (£142, 95% CI –£70 to £354), the total NHS costs at 12 months were higher in CORKA (£77, 95% CI –£138 to £291). By contrast, costs associated with private health-care use (–£15, 95% CI –£76 to £46), informal care (–£23, 95% CI –£210 to £164) and time away from paid employment (–£355, 95% CI –£820 to £110) were lower in the CORKA trial arm than usual care at 12 months. As a result, total societal costs (adding health care and other costs) were lower for CORKA than for usual care (–£316, 95% CI –£892 to £260).

Adopting an NHS health and social care perspective, the ICER for the CORKA home-based intervention versus usual care was £28,372, which is close to the standard threshold for cost-effectiveness in the UK. Adopting a societal perspective, the CORKA home-based intervention was cost-saving and more effective than, and thus dominant over, usual care.

Figure 11 presents the cost-effectiveness scatterplot that illustrates differences in mean total costs and QALYs for the CORKA home-based intervention versus usual care, adopting the NHS health care and social care (blue dots) and societal (red dots) perspective. Figure 12 presents the cost-effectiveness acceptability curve that gives the probability that the CORKA home-based intervention is cost-effective compared with usual care for different threshold values for a QALY (from £0 to £50,000 per QALY) for each costing perspective. Adopting an NHS or a health-care cost perspective, the probability that the CORKA home-based intervention is cost-effective is around 25% when the willingness to pay for an additional QALY is £0. As the willingness to pay for an additional QALY increases the probability of cost-effective increases, reaching around 43% at a threshold value of £30,000 per QALY. Adopting a societal perspective, the probability that the CORKA home-based intervention is cost-effective decreases with a higher willingness-to-pay threshold. The probability of cost-effectiveness is 75% at a threshold of £30,000 per QALY.

FIGURE 11.

Cost-effective scatterplot for the base-case analysis, assuming equal implant device costs.

FIGURE 12.

Cost-effectiveness acceptability curves.

Sensitivity analysis

Table 26 presents the complete-case analysis results at 1 year. Data were contributed to the complete-case analysis by 489 (79%) participants (usual care, 78%; CORKA home-based intervention, 80%). The results of the complete-case analysis were similar to those from the multiple imputation analysis. There was no evidence of a difference in QALYs by treatment allocation. NHS costs were higher and societal costs lower, although not significantly so, for the CORKA home-based intervention versus usual care. Adopting an NHS health and personal social care perspective, the CORKA home-based intervention was more costly and less effective than usual care and, thus, dominated by usual care. Adopting a societal perspective, the ICER for the CORKA home-based intervention versus usual care was £224,210, which is considerably higher than the standard threshold for cost-effectiveness in the UK. As usual care was more effective, the CORKA home-based intervention was, therefore, cost-effective, as higher ICER values indicate greater cost-effectiveness for CORKA.

Introduction and objective of the study

This qualitative study was carried out to explore the experiences of the patients participating in the CORKA home-based intervention and the physiotherapists and rehabilitation assistants delivering the intervention. We aimed to explore barriers to and facilitators of adherence and gain insight into experiences that might influence trial outcomes.

Methods

Findings

Physiotherapists and rehabilitation assistants

Using thematic analysis, we developed seven themes that cut across participants (Figure 13). We illustrate these ideas below with narrative exemplars.

FIGURE 13.

Physiotherapist and rehabilitation assistant themes.

Seeing the person in their world

This theme describes the value of seeing the person as a whole, not just a body part, in their own world. The home setting enhanced a holistic approach, rather than a biomedical approach:

I think that holistic approach is really, really important, so yes we’re there to get their knee bending . . . but in the bigger picture, I want them to be able to go outside and use their knee . . . to be able to go and see friends, or kind of do dancing and things like that . . . [In outpatients] it would have been very much focused on the knee.

Physiotherapist

Clinicians compared the clinical encounter, which tended to focus on the body part, with the more relational home encounter:

[Being at home] sort of takes it out of that context where the mindset of you’re just there just to fix you . . . you’re with that chance to be able to talk with people, they’re able to see you as being human . . . they were a lot more relaxed . . . They’re more open . . . they may well talk about and what their social lives are like . . . It’s a more open environment.

Assistant

In the home environment, the balance of power shifts. Entering a person’s home was described as a privilege in which you see the patient on their own terms: it is less paternalistic and the patient becomes a person:

It was very much on the patients’ terms whereas I can imagine kind of in a hospital setting, it’s a bit more kind of like on our terms.

Assistant

You’re in there and it’s a privilege to be in someone’s home. They’ve let you in. You’re not just seeing them; you’re seeing everything really . . . it’s quite invasive into their world.

Physiotherapist

Clinical staff reflected on the benefits of seeing the person in their environment and the changes that they would make to their future clinical practice to ‘see the person’:

I would say that I’d strive now to see the person as an individual . . . strive more to try and just understand where that person’s coming from, because even with something that you think is as simple as a knee replacement . . . in terms of the impact on that one person’s life . . . You don’t just know that stuff instinctively . . . you’ve gotta find out a bit about them, and so I think that’s hugely important.

Physiotherapist

Developing people skills

This theme describes the need to develop ‘people skills’ to effectively enter a person’s world. These skills are integral to effective treatment and develop with experience:

To be truthful they [clinicians] need to have good people skills to be able to talk to people and not talk down to them . . . don’t make ‘em feel like they’re on detention . . . it’s the way you go about it . . . personal skills count a lot . . . You’ve sort of got to win their confidence and once you do that . . . it’s amazing what you can get out of them once you get their confidence.

Assistant

Good ‘people skills’ meant that you had to get to know the person in their world and flexibly respond to their needs:

The reality is that things change quickly, you have to be pragmatic . . . real life is wonderfully, beautifully messy . . . you can’t know a thing that’s gonna come up.

Physiotherapist

Thinking outside the cubicle

This theme explores the way in which seeing a person in their own world fosters creative thinking and enhances positive outcomes. Clinical staff described the world in situ as the real world:

In a sort of sterile clinic or environment where the floor’s totally flat, there’s no obstacles . . . it doesn’t bear that much resemblance to somebody’s house . . . I think seeing people in their own home, it’s just different . . . being able to relate more to what the patient’s saying ‘cause you can see it, it’s not just a theoretical problem . . . they can show you.

Physiotherapist

Clinical staff enjoyed the freedom to be creative:

I enjoyed being able to give the people the realisation that you can do some exercise; you don’t need any fancy equipment . . . it’s sort of utilising the equipment that they’ve got, so their chairs, their stairs . . . to make use of, like, towels or, I don’t know, a bit of rope that the husband has got in the shed . . . It was taking it away from just ‘here’s a sheet with some exercises on it’.

Assistant

This creative approach was compared with the clinical space with its associated time constraints:

I think in the very time-pressured nature of outpatients, you probably can’t build up as much of a picture, and you can’t see the home environment that you’re working with, so it’s really great that we’re able to kind of walk through a day in the life of a, rather than just kind of talk through it . . . I think [we] probably get a bit trapped in the cubicle thinking.

Physiotherapist

I gained personally from doing that bit extra

This theme describes the personal gain from working holistically and recognising the impact that one can have:

I definitely gained a professional insight . . . but also personally as well: building those relationships, being able to gain more of an insight into human beings . . . this has been really, really important to me to be able to spend the time talking to people and understanding what makes them tick . . . I want to take on the stories; I want to learn more about people, (a) to help them, but (b) to kind of develop as well . . . personally and professionally . . . I take satisfaction from doing the little bit extra: from talking to the patient . . . I think if I wasn’t doing that, I wasn’t taking on the patients’ stories, I’d be less satisfied – and more likely to burn out . . . that’s the bit that I enjoy the, the above and beyond.

Physiotherapist

Clinical staff described the satisfaction that comes from having a positive impact on people’s lives:

I had this other chap, he lost his wife . . . I said to him, ‘What is your main goal?’ he said, ‘Walking to the grave’ . . . I said, ‘Well how far is it?’ he said, ‘Well I need to drive my car first’ . . . I said, ‘Well that’s not impossible . . . we’ll get there’ . . . I think it was about on the fifth time . . . he said, to me, ‘. . . I actually got to the cemetery . . . but I couldn’t quite do the walk.’. I said, ‘Well that’s an achievement, you must be really proud of yourself . . . and on the seventh time he said, ‘I’ve done it . . . I’ve actually gone to the grave’ . . . and I thought that was fantastic . . . it just feels such an achievement . . . It makes me feel so happy then.

Assistant

There is a fine line between patient and friend

This theme describes the challenge of managing the boundary between getting to know someone in their own world and remaining professional. These challenges were exacerbated in the home environment:

I think you always have to realise that the patient who you’re going out to see is a patient and not a friend . . . there is that fine line. I think it’s always being clear to them that you’re here to rehab them . . . I mean, some patients want to have a laugh and a joke with you but I wouldn’t say that’s unprofessional . . . that’s just making the patient feel comfortable . . . it’s polite to have that 5-minute conversation, how are you doing, how have you been, what have you been up to? And then once that’s out the way, get on with the treatment.

Assistant

Clinical staff described individual cases for which it was more challenging to remain professional. For example, there were those whom you ‘clicked with’:

They’re the patient and not friends. Some people you do sort of click with . . . so, I was almost like trying to be, sort of not friends, but be professional.

Assistant

I think the most tricky people are perhaps someone like, I’m treating a guy that’s a similar age to me . . . it’s just trying to maintain that you’re coming at things from an angle as a professional . . . maybe it’s more challenging when they’re, like, ‘Did you enjoy the pub the other night?’. [Laughter] . . . it’s again trusting [achieving] balance between the importance on knowing that person and building a relationship versus maintaining a professional identity.

Physiotherapist

Clinicians described a tension between having to develop good rapport to build their confidence and get the job complete and the need to maintain boundaries:

Building rapport is really, really important to get the patient engaged, but it’s managing the professional relationship is the difficult bit, I think . . . You do get to you know their family, you meet their kids, you know their dogs’ names . . . and you obviously share your own life stories as well . . . I found it kind of difficult to negotiate that barrier sometimes.

Physiotherapist

Clinicians described a ‘middle area of closeness’, in which you enter a person’s world far enough to be able to use your knowledge effectively, but not so far that you become ineffective:

There was one man [who] had other issues that he wanted to talk about . . . kind of brought [the assistant] into the drama of his neighbourhood. I think there is kind of a middle area of closeness . . . it’s not the more you get into their lives, the more help you can offer . . . there’s a threshold that helps, and then maybe as you get too far, you’re not objective . . .

Physiotherapist

Feeling outside my comfort zone

This theme describes times of discomfort for rehabilitation assistants working alone in a person’s home. Some described how they could feel vulnerable and on the spot as an unqualified health-care worker:

They were sort of under the impression that I was a physio[therapist], and then I say, “Well, I’m not a physio[therapist],” then they’re sort of, “What do you mean, you’re not a physio[therapist]?” . . . There was a couple of times where that’s sort of happened, and it just sort of made me feel a little bit, uncomfortable.

Assistant

Rehabilitation assistants described occasions where they asked for physiotherapist input, for example when patients exhibited severe pain, swelling or lack of expected progress, or when things just ‘didn’t look right’:

It was just mainly with the people who weren’t progressing as much . . . the lady who wasn’t getting beyond 30 degrees of movement . . . I felt I’d sort of done everything that I could. So, of course, I’d go back and touch base with [the physiotherapist] . . . even [the physiotherapist] was a bit baffled . . . he said, ‘Well . . . you’ve gone by the book . . . she’s doing what she can, she’s not progressing – she needs the MUA [manipulation under anaesthetic]’.

Assistant

Physiotherapists recognised that this sort of situation could be extremely challenging for rehabilitation assistants and emphasised the need for clear lines of communication:

I think the important thing would be just to emphasise openness and communication . . . ‘Don’t sit on something if you’re worried about something, tell me . . . I might be worried about it as well, and I’ll tell someone else’. [Laughter] . . . that wasn’t necessarily a formalised part of the training in the same way that we laid out how to deal with the exercises . . . but actually, probably just as important.

Physiotherapist

Needing a support network

This theme describes the need for peer support and effective two-way communication. Assistants described an effective support network in which someone had ‘tabs on them’; they did not feel abandoned:

I worked very closely with [the physiotherapist] . . . we did have a patient that was very tearful and rehab was quite difficult, and I called [the physiotherapist] as soon as I finished the appointment and we had a chat on the telephone and the next morning straightaway, we had a meeting to discuss how it was going on . . . which I think was really helpful because sometimes you just need that qualified [person] there just to cast their eye over it and just make sure everything’s going OK.

Assistant

This ongoing support was not only important on a personal level for the assistant, but also important in maintaining the patient’s confidence in the rehabilitation assistant:

If I ever [felt worried] I wouldn’t do it. Or I would get someone to come . . . I would always say, if you’re not sure of something just say . . . ‘I will find out’ . . . because it’s hard to change it when you go next time . . . then they could lose their confidence in you . . . Don’t put myself in a position where I say . . . ‘do that, do that’ and think, ‘oh my god no that could be wrong’.

Assistant

Physiotherapists described the challenge of balancing the need for support and fostering an environment of independence. Assistants described the need to feel trusted, yet also supported:

It’s quite nice from a physio[therapist] assistant point of view to feel like you’re making an impact independently and [the physiotherapist] trust[s] you . . . But at the same time, it’s nice to have that support . . . I didn’t feel at all like I was kind of abandoned or deserted . . . I think if the physio[therapist] was coming in every single time they’d be pressure . . . are they kind of judging [me]? . . . it was a really nice balance.

Assistant

This collaborative partnership between physiotherapist and rehabilitation assistant provided a safe place to learn new skills:

So, it is nice to have that bit of encouragement from a physio[therapist] . . . I took on a lot of skills and I learnt a lot about myself, I learnt a lot more about how to present myself to a patient because I was independent and I was learning from the [physiotherapists] . . . so I was pitching the best skills from everybody and putting them into, to what I want to become, so that was really good.

Assistant

Patients

We aimed to explore patients’ experience of being part of the CORKA trial. Using thematic analysis we developed themes that fell into three main categories:

1. the benefits of having treatment at home

2. the challenges of CORKA

3. factors that might have an impact on the treatment outcome.

The benefits of having treatment at home

Participants described five main benefits: it was a relief not travel; I got an hour’s work done in an hour; they can work with your surroundings; I wouldn’t have done it on my own; and there is nothing like company (Figure 14).

FIGURE 14.

Participant themes: the benefits of home therapy.

It was a relief not to travel

Participants described the difficulty that they would have in getting to the hospital and the great relief that it was to have been allocated to the home exercises. Often they would rely on other people to get them to the hospital:

I would have had to ask people to take me . . . we were having really bad traffic works at the time and it could take anything up to 2 or 3 hours to get in . . . you never knew how long it was gonna take you and the thought of having to make a physio[therapy] appointment . . . you’ve got somebody to take you and they’ve got to wait around and bring you back again . . . It’s exhausting doing those exercises . . . and then having to sit in the car and drive back . . . sitting in that car is exhausting when you got a bad knee, I can’t tell you, you’d never believe it, you really wouldn’t.

Female 29

It was a relief [to have treatment at home] . . . we’re fairly au fait with the bus services now but at the time I hadn’t been out so it was very much a sort of case of whenever I went anywhere it was calling a taxi to get there . . . because my partner has health problems as well, so what with that and the eye problems, and that we’re sort of permanently backwards and forwards.

Female 32

I got an hour’s work done in an hour

Participants felt that doing exercise at home made the best use of time, and described health-care experiences where their time had been wasted:

We did an hour’s work and we got on with it . . . it just worked all the time and that was good . . . I got an hour’s work done . . . [in hospital] I should think I got quarter of an hour’s work done in an hour . . . I prefer just to get on with the work and get it done with . . . it was just being kept waiting . . . it seemed to drag out a long time, over an hour, and you’d get very little done . . . you’re standing there thinking, ‘Come on, let’s get on with the next bit . . . I’m wasting my time here’.

Female 31

One woman described the need to consider the time cost to the patient and their employers (and family), not just the health-care costs:

I think sometimes it’s a matter of looking at the larger picture, because it’s not just the cost to the NHS it’s the cost to everybody isn’t it . . . you have to cost what it’s costing you for sending someone to the home, what it’s costing the individual, what it’s costing their employers . . . if you’re taking time off work to go to a hospital appointment and then you’re spending half the day there for what is essentially a 10-minute or a half-hour session . . . I know it’s expensive for the hospitals but it’s expensive for the patients as well . . . it is normally half a day isn’t it, instead of maybe half an hour of your appointment at home.

Female 32

They can work with your surroundings

Participants described the benefits of being in their own environment, as opposed to a ‘sterile room’. At home, the therapist can see what they have to work with and can improvise and adjust exercises; they can be creative and adaptive to individual goals in a real-life setting:

I have always maintained that physio[therapy] is best given in your own home because then people can see what you’re working with . . . [my] house is small . . . and that’s quite hard to explain when you say, ‘Oh, I live in a little house.’. People think, ‘Oh well, she can still walk around in a figure of 8,’ well you can’t, cos there isn’t enough room? . . . They can actually see what you’ve got and what you’re coping with . . . we went for a walk one day with the dog . . . huge confidence boost that was . . . they came here and looked at everything . . . I could see them looking . . . they’ve got to see what you’ve got and what you get round . . . you’re working with your surroundings, that’s why it worked so successful and it was such a good idea because . . . you were working with what that person has to live with every day.

Female 29

. . . knowing what your home environment is like and what you’re going to be able to do of those exercises . . . if they give you something sort of running for 10 yards or something inside, that’s not gonna work if you’re in a confined area, or if you can’t manage these stairs or you know if you’ve got nothing, no space, I mean we’re fairly compact here, there’s not a large area, just seeing that I think, and how you can manage the exercises in your home environment . . . Rather than . . . you get home and you think, ‘Well, how do I do that?’.

Female 32

I wouldn’t have done it on my own

This describes the confidence gained by being overseen by a health professional who tailors the exercises to the individual. Patients did not feel that they had the capability to progress in this way on their own:

They check you as to whether you were able to do this exercise or not and so . . . they were able to actually tailor the exercises to the way you were progressing. . . . now I wouldn’t have been able to do that on my own . . . [they] came in a couple of times as well and check that my knees were, my knee was flexible enough you know . . . without the contact there isn’t the progress.

Male 33

I found [exercises at home] very comforting because he was there and he was telling me what it’s for and what, what’s the best way of doing it . . . I sort of felt . . . confident because they, they were on-hand all the time.

Male 30

Participants described how knowing that someone would be coming to ‘check-up’ on them was a motivation to keep doing the exercises, and this motivation got them into the habit. There was a sense of duty towards the therapist who had come ‘all that way’; they did not want to let the therapist down:

It was, it was beneficial because I knew he was coming . . . which made me do the exercises . . . that was the main thing actually it kept me going . . . I feel that I’m so active because he came here . . . because he came, I felt I had to do them because I didn’t want to let him down coming all this way . . . I wouldn’t be as active. I am convinced that I would not be . . . ‘cos I am a lazy person . . . I probably wouldn’t be where I am now . . . whereas because I knew [name] was coming, it made me do them. You know, it’s as simple as that really.

Female 25

I think if he hadn’t have come I wouldn’t have bothered doing it . . . there was somebody there to make you do it too . . . which of course got me into the habit of doing it . . . I think if I’d just been given the sheets I would have thought, ‘Yeah, that’s alright, I’ll do that when I’ve got a spare minute,’ [laughs] . . . I mean I was a bit miffed at the time and thought, ‘Oh blow, I got to do that’ . . . once you get into a routine it’s not so bad, it’s just part of the day then, your ritual . . . I think we’re all people of habit aren’t we?

Female 31

There is nothing like company

Finally, participants describe the social benefits of a therapist coming to their own house. This social aspect facilitated exercise:

[I was] quite pleased to see her when she came . . . If you’re a bit younger and more active perhaps you might think differently, but in my position I was glad to see someone, I’m glad to see anyone that breaks the day up a little bit . . . for someone of my age and my way of life now, it was a welcome break to have someone come round and chat to you for 10 minutes, quarter of an hour. Umm, it’s got a bit of a social benefit I think.

Male 28

It did two things to me in my circumstances: it meant I had someone to talk to . . . So that was good, and not only that . . . I quite look forward to someone coming in . . . if you’ve got two people together one encourages the other . . . I mean you can go to these sports centres and . . . but I don’t think that’s the answer . . . it’s team sports you need isn’t it? . . . It’s also the social side that goes with it, isn’t it? . . . There’s nothing like company . . . I can vouch for that, there’s nothing like company.

Male 33

Challenges of having treatment at home

Participants also described two challenges of having treatment at home: they put me through my paces and feeling at sea now that it is over.

They put me through my paces

Participants felt a bit surprised about the pace and duration of exercises and felt exhausted:

Some days if I had gone through all the exercises I would have been doing them 24 hours a day . . . if I’d have been a sort of 40-year-old I would probably have coped with them, but being an 80-year-old was slightly different and I don’t know whether that was really taken into consideration.

Male 30

I was surprised how intensive it was . . . he did put me through my paces . . . pulling my leg up and stretching it out, that kind of thing . . . that really hurt and . . . the ones I’m doing is the lifting one which I find quite hard to do on this chair . . . and on the bed I can lift my leg up quite nicely and the bend of course, the bloody bend . . . which isn’t bad but it’s not brilliant, but it does hurt.

Female 25

I do feel a bit a sea now it’s over

Participants also described being worried about what would follow: they felt a bit abandoned. There was a sense that something should follow on, even if they had to organise it themselves:

I do feel a bit at sea, I’m thinking, ‘Hmm.’. And I know it’s gonna be up to me and sometimes I can think, ‘Oh’. But I think I need to see if I can find some . . . NHS physio[therapy] . . . Quite how I go about that I haven’t got a clue but I’ll think about it when I got my head . . . my head’s not quite in the right place yet . . . it’s something else you got to think about and organise.

Female 29

I done my 8 weeks or 6 weeks . . . and I think there’s a thing that, you can get a referral to the gym down in the town . . . I don’t know, I’ll have to make enquiries . . . who I’d have to get referred by . . . so, I don’t see them again then? . . . I want to know what’s going on . . . you get a lot of support in that first 6 months, but then there doesn’t seem to be anything . . . you feel as if right you’ve had your 6 months – bang that’s the end of it.

Male 30

Factors that might have effect outcome

Participants describe factors that might have an impact on their outcome from treatment: my get up and go has got up and gone; a variable social life; if you take to the person it goes a long way; and a lot would shirk from putting on a pair of running shoes.

If you take to the person it goes a long way

Participants described an effective health-care interaction as hinging on ‘taking to the person’. Personality factors were described as integral to being good health-care professionals and none of the participants was negative about being treated by an assistant therapist:

He’s [not a physiotherapist] bless him . . . but he’s also had injuries himself so he knows, he knows what it’s like to have a lot of pain . . . he’s an extraordinarily empathetic young man.

Female 29

A therapist should be kind, understanding, empathetic and be able to put people at ease. They should also be able to have a bit of fun and to know what ‘makes people tick’. They should not appear disinterested or ‘perfunctory’:

[Previous treatment] was perfunctory . . . it was done by the book . . . they’ve got to tick certain boxes . . . a lot of physio[therapy] workis personality . . . you can log-on to that personality with someone, you’ll get far more out of them . . . a lot of physio[therapy] is how you interact with people . . . it was a personal thing . . . interpersonal skills are just crucial. You’ve got to be able to talk to people on any level really . . . got to be able to relate to what that person is telling them . . . you have to have empathy . . . you have to be able to cotton on to what makes them tick really . . . the best ones put an element of fun, they get on your wave length you know, pretty quickly . . . It’s just building up a relationship . . . you have to do it very quickly don’t you . . . so you have to have the innate ability to build up a fairly good relationship with somebody.

Female 29

[He] was wonderful, he really was . . . everybody has been very, very kind and all gone over the top with me, let's put it that way if you like, with the helpfulness . . . I can’t fault anything. I really can’t. . . . I liked [name], you know, she kind of put me at my ease, never put any pressure on me at all . . . I think if you take to the person that’s talking to you it goes a long way to anything, you know.

Female 25

One woman made a comparison to palliative care therapists:

I used to listen to [palliative care] physio[therapist]s all day, getting people to go up and down these stairs, who didn’t want to . . . they were dying, they didn’t want to know about it but she was gonna get them up. And the way she did it was amazing, because, because of her personality . . . cos it’s not about dying, it’s about making the most of living really . . . I think they’re quite similar actually . . . the empathy and the relating to people . . . I mean, you’ve got to have all those qualities haven’t you? You’ve got to have the same qualities.

Female 29

My get up and go has got up and gone

Participants describe other health issues with advancing age that might have an impact on recovery, including high blood pressure, chronic arthritis, hip replacements, diabetes, heart attacks, itching, chronic cough, hearing loss, sight loss and incontinence. Some described this as ‘just old age’ or ‘my get up and go has got up and gone’:

I developed diabetes . . . then I got angina 3 years later, I had a heart attack . . . now I’ve got a terrible blasted itch I’ve had it for 4 years, no one knows what it is. . . . It makes my life an absolute misery. . . . Another thorn of mine: hearing, I’m completely deaf in this ear . . . and I’ve had this cough since about December I suppose but that’s the least of my problems. I think most of my troubles are just age . . . Lack of stamina compared to when I was young . . . put it another way: my get up and go has got up and gone . . . I mean age has to catch up with everyone doesn’t it eventually, unless you’re gonna die in your sleep at 50, that’s luck . . . I can’t die young can I now? I think that’s really the basic trouble: age. I think so.

Male 28

I do have other health problems . . . I used to enjoy swimming a lot and I’ve not been able to do that for about 5 years . . . that was partly my knee but . . . it’s the eyesight issue as much as anything . . . vision certainly affects what I decide to do.

Female 32

A variable social life

Some participants described a diminished social life that could have a profound impact on their motivation and recovery. Two men had quite recently lost their wives and described the impact of this loss:

I think my mental thing is, is more related to the loss of my wife in many ways . . . Because I’m in a house on my own . . . I don’t have anybody to talk to . . . I get up in the morning to have my breakfast . . . but some days I don’t bother to have any food at all, because I can’t be bothered. We always used to lay the table up here for lunch, you know, and we would sit down and have a chat; television was turned off, radio was turned off, we used to talk at lunchtimes always . . . that’s the big issue with me mentally . . . Now whether that’s had an impact on the way I’ve, I’ve recovered I don’t know . . . but I feel it has had an effect . . . I still talk to my wife [laughs].

Male 33

I don’t do much really. Umm, my daughter has every Friday off and does the housework and this that and the other, she looks after me quite a lot when she’s here, but she’s not here a great deal. Umm, I don’t get out a lot at all . . . I watch the television mostly in the evening; I haven’t succumbed to daytime television yet . . . all my friends have died really . . . And you do lose a lot of friends when you stop work. But once again they’ve all died off one by one; almost the longest survivor I think.

Male 28

Others described a more active participation in life with friends and family:

I play Bridge a lot . . . I teach Bridge . . . I enjoy doing that . . . they wanted to learn over at the club . . . So I said to the girls, ‘Do you want me to do it?’. ‘Oh yes please’ sort of thing, and so I, I started them from scratch and that was 14 years ago . . . and we’re still learning [laughs]. Yes, once a week. And then I’d go with one or two of them to different clubs, a couple of clubs to get them used to going out and joining clubs and that, yeah they enjoy it, it’s nice.

Female 31

I have two sons and a daughter and that family with the grandchildren and their partners and I’ve begged one of them to give me a wedding [laughter] . . . they’ve all got partners, two step grandchildren, great grandchildren . . . they’ve been together 10 years and they’re just expecting their first child together . . . I'm very proud of them all, they've all got homes, jobs and if they’re not buying a home they're renting one.

Female 25

A lot would shirk putting on a pair of running shoes

Personal motivation to exercise was described as a factor that might influence adherence and, therefore, outcome. Participants described the ‘exercise type’ or ‘not the exercise type’:

I do have to give myself a good talking to . . . you’re talking to the wrong person, cos I am literally the laziest person, I hate exercise. Yes. And you know I have a lovely place to walk in, you know, I used to say that to myself before, ‘I don’t live in a tower block where I’ve got to walk round and round the sofa’, you know I just have to step out the door and this time of year it’s beautiful, you know. And I don’t even have to go very far, I can just walk down the village and stop for a cup of tea down the other end and come back again. You know, I have no excuses other than the fact that I am just bone idle and I hate exercise.

Female 29

They’ve probably never exercised in the first place . . . a lot of individuals that I know would shirk putting on a pair of running shoes [laughs] . . . My wife was a townie . . . She’s never been interested in sport. She’s never done any sport at all in her life . . . When I was at school we had . . . form games . . . school games . . . and then you had the school represented teams . . . so I’ve always been physical as it were. But I mean my son he’s, he, he’s not physical, he doesn’t do exercises, he would blow a gasket if you asked him to do some exercises.

Male 33

Discussion

Introduction

This chapter describes the development of the screening tool to identify participants at risk of a poor outcome after TKA, who would then be suitable for recruitment into the CORKA trial. During trial design, we identified that no existing screening tool existed to identify patients at risk of a poor outcome after KA. It is reported following the Transparent Reporting of Multivariable Prediction Models for Individual Prognosis or Diagnosis statement. 105,106

Study population

The data used to develop the screening tool came from the Knee Arthroplasty Trial (KAT) data set. 107 The KAT was a pragmatic, multicentre RCT designed to determine whether or not a metal-backed plate for the tibial component was more effective and cost-effective than a single high-density polyethylene component or resurfacing the patella, and whether or not a mobile bearing between the tibial and the femoral components was associated with better outcomes than standard designs without a mobile bearing. 107

The KAT involved 116 surgeons at 34 UK sites. All patients under the care of a collaborating surgeon were potentially eligible for inclusion in the trial if a decision had been made to have primary knee replacement surgery. A patient was ineligible for the trial if their surgeon considered that a particular type of operation was clearly indicated, for example if a patient required a highly constrained knee replacement to replace function of the collateral ligaments. A participant remained eligible only if their surgeon was convinced that there was no indication for one of the trial surgeries. For example, a patient with a very thin patella would not be eligible for the patellar resurfacing comparison because patellar resurfacing would not be suitable for them. 108

Knee Arthroplasty Trial data collection

The KAT data collection was carried out using standard forms to record pre-operative, peri-operative and post-operative information. Data describing functional status and QoL were collected in forms that were sent to the participants by mail. These follow-up forms were meant to be completed by participants around 3 months, 1 year and 2 years after the surgery.

If follow-up questionnaires were not returned after a mail reminder, the trial team gave a telephone reminder and offered participants the option of completing the form over the telephone. Further details of the KAT design and data collection can be found elsewhere. 107

The KAT time points and the data collected at each are described in Table 27.

Individual patient data used to develop the screening tool

The KAT data set contained individual participant information at baseline and 3-month, 1-year and 2-year follow-up, which was collected from 2318 participants randomised to receive one of the three KAT interventions.

To develop the CORKA screening tool, we excluded data from KAT participants aged ≤ 55 years (n = 115) to match the CORKA trial exclusion criteria. We also excluded data from KAT participants with a BMI of > 50kg/m² (n = 9). The final development data set comprised information from 2194 KAT participants.

Available variables and initial selection of candidate predictors

Of the variables available in the KAT data set, those measuring sociodemographic indicators (e.g. age, gender and BMI), pre-surgery QoL, physical function, current mobility levels and pain status were considered potentially useful for identifying KA patients at risk of a poor surgery outcome. We selected eight of these variables as a priori candidate predictors for testing for inclusion in the screening tool. These candidate predictors were selected based on their clinical importance, existing statistical evidence and relevance to the patient. The initial selection was made internally by the research team, taking into account the literature and the team’s expertise.

Table 28 lists the pre-selected candidate predictor variables, what type of variable they are, how they were collected and the number of missing data. The table also includes information for height and weight, which were used to calculate the eighth variable, BMI.

Outcome definition

The outcome of interest for the predictive tool was functional status 1 year after TKA, measured by the OKS. To identify the participants who would most benefit from the CORKA home-based intervention, we defined an OKS score of 26 or less as the threshold for a poor outcome.

The OKS was the primary outcome measure in KAT. Table 29 presents a summary of the available outcome data in the KAT data set, the number of participants with a good or poor outcome and the number of missing data.

Exploratory analysis and data transformation

The baseline characteristics of the candidate predictors in the KAT data set were summarised using means and SDs for continuous variables, and frequencies and percentages for binary and categorical variables.

Binary and categorical predictors were tabulated against the outcomes to check for empty or low cell counts (n < 5). Where empty or low counts were found, categorical variables were re-categorised by joining some of the categories together if it made clinical sense to do so.

Although most of the KAT participants had their Short Form questionnaire-12 items (SF-12) assessment performed with version 2.0 of the questionnaire, some were assessed with version 1.0. The two predictors that were based on questions from this instrument were re-categorised to ensure data consistency.

The list of manipulated variables and the changes made for each are presented in Table 30.

Handling missing data

The percentage of missing data in the CORKA screening tool development data set was presented for each candidate predictor variable in Table 28 and the outcome in Table 29. To conform to current guidelines, multiple imputation for all participants with at least one missing value was performed. 109 As several predictor variables of different types had missing data (binary, categorical and continuous), multiple imputation by chained equations was carried out using the mi impute chained function in Stata 15. We used the option logit for binary variables, mlogit for categorical variables and truncreg for continuous variables, setting the lower and upper limits for imputed values as the correspondent scale limits.

We assumed that all missing data were missing at random and that the imputation models included all available observed characteristics for the predictors of interest, predictors of predictors (e.g. weight and height for BMI) and outcomes, as recommended by White et al. 109 There were low rates of missing data, with no more than 6% missing for any one candidate predictor. Based on the number of missing data for the variable with the highest rate of missing observations, we produced 10 complete imputed data sets. We did not perform data transformations on continuous predictor variables before imputing missing observations.

Despite using the augmented-regression approach,110 some categorical predictors were excluded during imputation owing to perfect prediction. 111 Perfect prediction occurs when one of the categories in a categorical explanatory variable is always observed with one of the possible outcomes, for example if all women had a good outcome. Perfect prediction is a problem because it leads to infinite coefficients with infinite standard errors, causing instability during estimation and preventing the imputation model from achieving convergence. It is often resolved by discarding the observations corresponding to the offending covariate patterns or by discarding the independent variables that perfectly predict the outcome during estimation. We decided to keep all of the observations in the data set and discard the variables that caused perfect prediction from the multiple imputation model. As a result, we discarded two baseline variables: ‘mobility’ and ‘have you felt downhearted or low’.

Sample size considerations

Sample size requirements for logistic regression are based on the concept of events per variable (EPV). It is widely recommended that a data set contains a minimum of 10 EPV to avoid overfitting the model. 112–117

As the KAT database contained 389 events of a poor outcome 1 year after surgery (see Table 30), we could examine up to 39 (389/10) candidate predictor variables in the model. After excluding two of the preselected candidate predictors owing to perfect prediction, only six of our baseline candidate predictors remained for testing. We, therefore, had 65 EPV available, making overfitting unlikely.

Data modelling

As the outcome was binary (poor outcome after TKA: yes/no), the screening tool was developed using a logistic regression modelling framework with the logit probability of an adverse outcome as the response variable. We removed candidate predictors that were not statistically significantly associated with the outcome. The remaining candidate predictors were included in the final logistic regression. As we had 65 EPV, we did not shrink the regression coefficients.

To aid implementation of the screening tool, we simplified the final multivariable model by assigning integers to the predictors based on the final model. The integers can be summed to obtain a total to identify participants who are likely to have a poor outcome 1 year after TKA.

Predictive accuracy

The performance of the CORKA screening tool was characterised by evaluating its discrimination. Calibration was not assessed, as the aim of the screening tool was to generate a score for an individual and not predict the probability of the outcome.

Discrimination measures the screening tool’s ability to correctly rank individuals. The overall discriminatory ability was summarised by the area under the receiver operating characteristic (AUROC) curve (c-index), with 95% CIs. The AUROC was classified as follows: 0.5 to < 0.6 poor, 0.6 to < 0.7 fair, 0.7 to < 0.8 moderate, 0.8 to < 0.9 good and 0.9–1 excellent. Owing to the large number of EPV, we did not carry out a formal internal validation of the model using bootstrapping or cross-validation, for example.

Results

Table 31 summarises the candidate predictor data available for modelling in the KAT data set before and after exclusions and multiple imputation. Multiple imputation and excluding KAT participants with characteristics that would not meet the inclusion and exclusion criteria for the CORKA trial from the model development data set did not greatly alter the sample’s characteristics. For example, average age changed from approximately 69.99 ± 8.38 years before to 71.01 ± 7.22 years after exclusion and multiple imputation, and BMI changed from 29.70 ± 5.47 kg/m² before to 29.53 ± 4.88 kg/m² after exclusion and multiple imputation.

We kept continuous variables in their original format without data transformation for multiple imputation of missing data.

A summary of the unadjusted and fully adjusted multivariable models’ estimates (odds ratios with 95% CIs and p-values) is presented in Table 32. In the unadjusted analysis, all remaining candidate predictors were significantly associated with the outcome, except for gender. In the fully adjusted multivariable model with the predictors, all variables remained statistically significant. Age was negatively associated with the likelihood of a poor outcome 1 year after TKA. The research team made the pragmatic decision to omit this variable from the final multivariable model, as we did not have a plausible clinical explanation for the observed direction of association.

A summary of the final multivariable logistic regression model (odds ratios, 95% CIs and p-values) is presented in Table 33.

As all of the odds ratios were greater than 1, the CORKA screening tool was developed by rounding the observed odds ratios for each predictor or predictor category. Any patient with a score of 5 or more would be classified as being at increased risk of a poor outcome after 1 year and likely to benefit from the CORKA home-based intervention. The threshold score of 5 was a pragmatic choice taken by the research team based on forecasted numbers of individuals declared as at high risk of a poor outcome. The final CORKA screening tool is presented in Table 34.

The model performance was assessed in terms of discrimination. The model had a fair overall discriminatory ability to predict a poor outcome 1 year after TKA (apparent performance), as measured by the AUROC curve (0.66, 95% CI 0.64 to 0.69).

Discussion

The CORKA screening tool was developed to be a simple-to-use tool that identified participants likely to be at increased risk of a poor outcome 1 year after TKA and, therefore, most likely to benefit from the CORKA home-based intervention. As the screening tool was developed using a large existing data set with a small number of candidate predictors, we were not concerned about overfitting the model. The apparent performance of the model was fair, as measured by its discrimination.

The research team was confident that the developed screening tool would help to identify most patients at an increased risk of a poor outcome and most likely to benefit from the CORKA home-based intervention, owing to the quality of the development data set, data optimisation with missing data multiple imputation and model building strategy.

We did not conduct any cross-validation or internal validation because EPV is a marker of the potential for overfitting, with an EPV value of < 10 widely considered to be the level at which concerns may be raised.

Our EPV for model development was very large at an EPV of 65. Simulation studies118,119 illustrate that once EPV is > 20, the bias owing to any overfitting is negligible/eliminated, such that the apparent performance (reported in the monograph) converges to the true underlying large sample performance. 118,119

We chose to use the KAT data set and the OKS as an indicator of poor outcome because of the availability of a large data set. The use of clinical trial data is potentially challenging, particularly if there is a large treatment effect observed in the trial. The potential of using clinical trial data has recently been described by Pajouheshnia et al. 120 Options are to include treatment (if the effect is large) in the model or to use the control arm only (particularly if the arm was a placebo arm). The KAT observed no differences in OKS between the treatment arms, which reinforced our justification of using the entire trial data. Furthermore, available data following the natural course of patients are uncommon, individuals will usually have received specific treatments during follow-up, i.e. confounding by indication. 120

Overview of the trial findings and key messages

Knee arthroplasty is one of the most commonly performed musculoskeletal operations. Although it effectively reduces pain and improves physical function for most patients and has good reported cost-effectiveness, around 15% of patients report dissatisfaction with their post-operative outcome. 7,121,122

In the short term, KA results in substantial loss of quadricep strength and decreased range of knee motion and function, compared with the pre-operative status. 123 Pain, reduced function, decreased mobility and resulting falls cause significant morbidity for those patients with a poor KA outcome. However, there is evidence to support the use of functional exercises by physiotherapists to improve the range of joint motion and QoL. 18 Draft clinical guidelines from the National Institute for Health and Care Excellence (NICE) recommended targeted physiotherapy interventions for patients at higher risk of a poor outcome for whom a self-managed exercise programme at home may not be suitable. 124

The CORKA trial was a large multicentre RCT that targeted patients who were at risk of poor outcomes after KA. We randomised 621 participants: 312 to usual care and 309 to home-based rehabilitation. Most participants had a screening tool score of 5 or 6, all met the criteria of moderate or severe knee pain and most were overweight or obese. The two treatment groups were well balanced at baseline. The average age of the trial participants was 70.5 years.

Approximately one-quarter of participants received a UKA. TKA has been the traditional treatment of choice for older patients with knee pain and limited activity because of arthritis. Undertaking UKA is possible for some of these patients, but the indications outlined by Kozinn and Scott125 in 1989 suggested that the procedure was not suitable for patients with higher degrees of comorbidity. The original indications for UKA included arthritis isolated to the medial compartment in patients under 60 years of age with low levels of physical activity and weighing under 82 kg. This was thought to be around 6–12% of patients undergoing knee replacement surgery. 126 However, more recently, Campi et al,126 reported that surgeons have widened the indications, especially for weight, age and activity, and up to 50% of patients may now be eligible for UKA.

The recent TOPKAT trial127 compared TKA with UKA, and demonstrated that both were effective options for patients with medial knee arthritis with similar clinical outcomes. However, participants receiving a UKA had slightly better outcomes in terms of lower surgery costs and follow-up health costs in the first 5 years. The lower costs of surgery for UKA may be a result of the increasing use of enhanced recovery pathways, which are having a positive effect on outcomes. 121 These pathways can adopt rehabilitation strategies that encourage early mobilisation and discharge. 128 Within the CORKA trial, the same post-operative intervention was used irrespective of implant. However, both treatment options may have been shorter and less intense for the UKA patients, who rarely struggle to regain their range of movement.

There was no difference in evidence of the benefit between usual care and home-based rehabilitation in terms of LLFDI function score at 12 months post randomisation (ITT multiple imputation difference 0.49, favouring CORKA; 95% CI –0.89 to 1.88; p = 0.48), 6 months post randomisation (ITT multiple imputation difference 0.66; 95% CI –0.70 to 2.03; p = 0.34) or based on any sensitivity analysis. As previously discussed, this may be in part because participants in the usual-care group received more treatment sessions than anticipated. Treatment effects varied substantially by site and showed some relation to the average number of sessions of usual care and home-based rehabilitation received at each site.

There were also no significant differences between the two groups in terms of any of the secondary PROMs or physical measures for the ITT population using available cases. Sensitivity analyses of the key secondary PROMs (LLFDI disability frequency and limitation, OKS and KOOS QoL subscale) identified one significant difference between the two groups, for LLFDI disability limitation using the ITT population and multiple imputation. However, we do not believe that this is a difference that is explicable by the package tested and think that it is more likely to arise from the number of tests performed than being a true finding when considered in the context of all the other tests, physical and self-reported, that were conducted with no statistically or clinically significant differences. The number of participants experiencing a serious adverse event was small in both the usual-care arm (14/312, 4.5%) and the home-based rehabilitation arm (18/309, 5.8%), with no significant difference between them (risk difference 1.3%, 95% CI –2.1% to 4.8%; p = 0.45).

Over the 1 year of the CORKA trial, we found a small non-significant difference in QALYs favouring the CORKA home-based intervention (0.003 QALYs, 95% CI –0.017 to 0.023 QALYs). The CORKA intervention itself was cheaper, but the participants in the CORKA arm reported higher use of other NHS resources in that year than participants in the usual-care arm. This cost was offset by lower societal costs, with more CORKA participants returning to the workplace and using less social services than usual-care participants.

The participant diaries that were used to capture health and societal costs showed that participants in the CORKA arm had less time off work and returned to work sooner than participants in the usual-care arm. We do not believe that this is a spurious finding. We can only speculate on the reason for this. The CORKA intervention was much more functionally focused, with individual goal-setting aiming to return to activities that the participants had selected as important to them. Conventional physiotherapy has a more biomechanical approach focusing on strength and mobility. We believe that the more individually tailored functional approach of the CORKA intervention may have been a factor in a larger number of people returning to work. Overall, however, because of the age of most participants, a relatively small number of participants in both groups returned to work.

Usual care used a traditional clinic-based or hospital-based model of outpatient physiotherapy delivered by qualified, registered physiotherapists. By contrast, the CORKA home-based intervention was multidisciplinary in content, was delivered in participants’ own homes and used a staffing model of rehabilitation assistants supervised by a qualified physiotherapist or occupational therapist. The trial data suggest that the two treatments were similarly effective and safe in this population.

Interviews with patients who received and therapists who delivered the CORKA home-based intervention showed that it was acceptable to both patients and clinicians. Being treated at home facilitated a holistic rather than biomedical approach that underpinned a collaborative partnership.

Internal validity and methodological limitations

The CORKA trial was pragmatic in the context of the funding envelope available for physiotherapy within the UK. It sought to estimate the effect of treatment strategies based on clinic-based or home-based treatment using different staffing models. It used a range of recognised measures with a 12-month follow-up, which is longer than many previous trials.

We recruited 621 participants. Based on our published statistical analysis plan,46 the sample size calculation required data from a minimum of 620 participants, assuming a moderately small standardised effect size of 0.275, or 3 points on the LLFDI function score, with a power of 90% and an alpha of 0.05. This allowed for a withdrawal rate of 10%.

The availability of follow-up data in this trial was very good. Primary outcome data were available for 92% of participants in the usual-care group and 89.4% of participants in the home-based rehabilitation group at 12-month follow-up. Rates of withdrawals (34/621, 5.5%) and deaths (3/621, 0.5%) were very low, resulting in a loss to follow-up rate lower than the 10% we had allowed for. We had primary outcome data at 12 months for 566 participants (91.1%), providing sufficient statistical power to detect a difference between the groups as originally specified.

Randomisation was conducted by a central facility at the Oxford Clinical Trials Research Unit and was stratified by site using variable block lengths, preventing research staff from anticipating or influencing treatment allocation for any given participant and resulting in two well-matched arms. Owing to stratification by recruiting site when allocating participants to the groups, the numbers recruited were not divided into exactly equally sized groups, but did not differ significantly. Demographic data, self-reported functional measures, clinical impairment measures and disease stability were all very similar.

Owing to the nature of physiotherapy interventions, it was not possible to blind participants to their treatment allocation. However, we endeavoured to ensure that outcome assessors remained blinded throughout the trial. This was hard to achieve, with a high risk of inadvertent unblinding by participants to research staff. We used a number of strategies to reduce the risk of unblinding, including explicitly instructing participants not to inform their research physiotherapist of their group allocation at every visit. 129 Barker et al. 130 have previously reported that using these strategies resulted in successfully maintaining blinding in 81–91% of the assessments, meeting the expected level for a successfully blinded trial, as defined by Minns Lowe et al. 129 and Boutron et al. 131 As the primary outcome measure was self-reported by the participants, it was unlikely to have been influenced by unblinding of the outcome assessors.

Compliance with home-based rehabilitation was prespecified as receipt of at least four treatment sessions, and the compliance rate was high (269/309, 87.1%). The average number of treatment sessions in each group was similar, with participants in the usual-care group receiving a median of four sessions (IQR two to six sessions) and participants in the home-based rehabilitation group receiving a median of five sessions (IQR four to seven sessions). The number of treatment sessions received in the usual-care group was larger than anticipated, which may have affected the observed treatment effect. There was also wide variability in the number of usual-care treatment sessions offered to participants (range 0–27). The median number of usual-care sessions differed substantially by site, ranging from two at site 13 to six at site 1. This suggests that there is a lack of standardisation in usual rehabilitation offered to these patients across the country.

The interventions in both of the treatment arms were well received by participants according to the qualitative study results.

Quality assurance checks and observation visits determined that the therapists delivered the content well, although there was variation in the number of sessions delivered.

The treatment logs and assurance visits demonstrated that exercise progression was delivered as planned. In designing the interventions, incremental progression of exercises and activity levels were considered important to achieve the underlying physiological changes required to improve strength and balance. There was good evidence from the content analysis of treatment logs that participants had progressed the quantity and difficulty of the programme content, incrementally increasing the amount of activity expected throughout the course of the treatment programme.

The trial team achieved a high level of data completeness for the primary outcome, with over 90% complete. The completeness for the physical measures was slightly lower at 89.2% for the F8WT, 89.4% for the 30SCST and 89.9% for the SLS, but was still close to the desired level of 90%, reflecting that the physical outcome tests required a visit to the site. The pragmatic approach to collecting outcome data by post or telephone when participants were unable or unwilling to attend face-to-face assessments resulted in a higher completion of the questionnaire data.

A small proportion of participants (5.8%) withdrew and were distributed evenly across the two treatment arms. There were no significant baseline differences between those retained and those lost to follow-up. Only three participants (0.5%) were lost to the trial because of death. This was lower than expected considering that, on average, participants were over 70 years of age and had a BMI of over 31, and 81% were categorised as being in an ASA classification category of mild or moderate systemic disease at baseline.

An internal pilot study was conducted at one site (Oxford) to review the recruitment feasibility and confirm the intervention package. The 15 participants randomised during this pilot study were used in the final analysis, as no changes were made to the interventions or assessment tools between the pilot and the main trial.

External validity/generalisability of findings

Overall, we believe that the generalisability of the findings of the trial is good. We recruited sites across a number of NHS trusts, resulting in variation in trust size and a range of site sizes and geographical locations. The training burden on qualified therapists and rehabilitation assistants was relatively small because the core principles underpinning the programme were already within the skill set and expertise of most NHS therapy staff. Thus, we believe that the intervention could be implemented in other NHS sites relatively easily, particularly where a model for community care already exists.

Critique of methods

We designed our trial to test interventions that could be delivered in routine NHS physiotherapy practice within the current commissioning constraints of the number of sessions remunerated by most care commissioning groups. The trial showed that it was practicable and successful to deliver the intervention content in both clinical and community settings.

Study limitations

There are limitations to our work. We developed a screening tool to identify the patients in the worst half for predicted outcome. To make this pragmatic, we based the key questions on those already collected as standard practice in most pre-operative assessment clinics or collected as part of the standard national PROMs data set for KA. This may have resulted in a tool that was insufficiently sensitive to identify the patients most at risk of a poor outcome that could be addressed by rehabilitation after surgery. In developing our screening tool, we made an assumption that participants with an OKS of 26 or less would be more likely to benefit from the home intervention than from traditional physiotherapy. We recognise that this assumption was informed by the consensus opinion of groups of orthopaedic physiotherapy practitioners that we consulted in developing our research study and interventions and that it is not self-evident or based on hard evidence.

As we compared the multidisciplinary home-based intervention with the current UK best practice of outpatient physiotherapy, we cannot comment on the effect of the treatment arms compared with no treatment. The research available to date focuses on the efficacy of different types of physiotherapy following KA, with a lack of consensus. However, physiotherapy remains an accepted component of the treatment pathway for KA in the UK. A control group withholding all post-operative outpatient physiotherapy was not possible, as it was judged unacceptable to surgeons (who would not allow their patients to enter into such a trial) and unethical to deny physiotherapy to patients already highlighted as being at risk of poor outcomes.

We chose the LLFDI as the primary outcome measure, which is not an outcome tool specifically designed to capture change in outcome after KA. It was selected as it maps to the World Health Organization’s International Classification of Functioning, Disability and Health, and allows analysis of function, activity and participation. It has also been used in other rehabilitation trials of older adults. It is arguable that a more specific measure targeted at KA may have been a better choice. We included the OKS as a secondary measure to address this potential limitation. The results of the LLFDI and OKS were well-matched.

Owing to the nature of this therapy trial, it was not feasible to blind patients or therapists to treatment allocation. To ensure generalisability, we compared physiotherapy interventions that are deliverable within the current standard commissioning envelope for NHS physiotherapy, setting physiotherapy provision to approximately six sessions of individual physiotherapy.

Our cost-effectiveness analysis has several limitations, including missing data on resource use and EQ-5D-5L. We accounted for missing data using multiple imputation, which assumed that the data were missing at random conditional on modelled covariates. 103 We found no strong evidence to contradict this assumption and found that multiple imputation and complete cases analysis gave the same results about the cost-effectiveness of the CORKA intervention.

Implications for clinical practice and policy

The NHS 5 Year Forward Plan156 has identified the need for further productivity and efficiency in health-care delivery. It has also identified that the demands on health services are growing owing to an ageing population with more long-term conditions and greater patient expectations. Part of the NHS plan is to decentralise services, with greater community provision and less emphasis on care provided in large acute trusts where that is not clinically mandated.

We suggest that the CORKA home-based intervention is a model that meets this strategy as it targets care to those at higher risk who most need it, within participants’ own homes and communities. It addresses the issue of workforce shortage by using an innovative workforce model of advanced rehabilitation assistants, moving UK service provision closer to that which has been proven to be effective in North America, where the use of physical therapy assistant graded staff is well embedded.

Patient and public involvement

We benefited from the support of highly engaged patient and public involvement (PPI) representatives at various stages of the trial. Before the trial began we were supported by the Biomedical Research Unit PPI officer, who helped in the recruitment of a PPI representative as a co-applicant and fully voting member of the CORKA Trial Steering Committee. This PPI representative had previously undergone KA surgery with a relatively poor outcome. They, therefore, understood the rationale behind the CORKA trial, in addition to the aim of the intervention to help those at risk of a poor outcome. The trial team were able to update the PPI member throughout the study, with feedback welcomed.

Patient and public involvement engagement also occurred in the development of the intervention used for the CORKA trial. In the development and piloting phase, changes were made to participant materials and intervention procedures based on comments and advice from people who had undergone KA surgery; these are covered in more detail in Chapter 3. More informally, we were extensively supported by members of our local research engagement group and by former and current patients undergoing physiotherapy following KA. Their generous time and comments were hugely influential in shaping the trial, designing the intervention, and producing the patient information and documentation to support the trial. The PPI representative supported the write up of the CORKA monograph by reading the chapters that they were involved in and PPI groups will be used to help disseminate the results of the CORKA trial.

Further research

The findings from the cost-effectiveness analysis are important and consistent with the findings of the trial: there is no meaningful difference in costs and QoL between the CORKA intervention and the usual-care arm at the 12-month follow-up time point, although there is some evidence that CORKA may be cost-effective from an NHS and societal perspective. Further work to review the relative importance of NHS and societal costs in this population may be important, particularly if health care moves towards a more centrally funded health and social care integrated model.

It is suggested that further research should focus on developing a screening tool that is more sensitive in identifying those patients who will benefit from additional input, as recommended by the draft NICE guidance. 124 The NICE committee did not feel that self-directed rehabilitation without further supervision was adequate, and did not recommend supervised rehabilitation for all. However, they commented that ‘it would be very useful if a tool existed that indicated those who would benefit from supervised rehabilitation or required adaptations to self-directed rehabilitation’ [reproduced with permission from NICE124 © NICE 2020 Joint replacement (primary): hip, knee and shoulder. Available from www.nice.org.uk/guidance/ng157 All rights reserved. Subject to Notice of rights NICE guidance is prepared for the National Health Service in England. All NICE guidance is subject to regular review and may be updated or withdrawn. NICE accepts no responsibility for the use of its content in this publication].

The screening tool developed in the CORKA trial identified the population most at risk of a poor outcome after KA. It could be further developed to identify factors linked to engagement with rehabilitation.

The CORKA home-based intervention was delivered by rehabilitation assistants supervised by qualified therapists in a ratio of five sessions to two sessions. To the best of our knowledge, the use of different workforce models and particularly a hybrid model of assistants supervised at a distance by qualified therapists has not been researched in any depth in the UK. With projected deficits in health-care staffing from NHS England, modelling further work looking at different workforce models and interventions using rehabilitation assistants may form an important subject for further research. The use of a different staffing model is also worthy of exploration in terms of overall costs and affordability, making good quality rehabilitation affordable to all, particularly with the anticipated continued growth in the number of KA procedures performed each year.

Research to explore the relative benefits of alternative ways of delivering interventions would also be useful, particularly if this improved adherence. Similarly, it would be of interest to explore the reasons that older people do not engage fully with treatment interventions to improve the delivery of future programmes using strategies to enhance engagement.

The CORKA trial hospitals

Ashford and St. Peter’s Hospital NHS Foundation Trust (Ashford Hospital), Countess of Chester Hospital NHS Foundation Trust (Countess of Chester Hospital), Dorset County Hospital NHS Foundation Trust (Dorset County Hospital), Dorset Healthcare University NHS Foundation Trust (Dorset Healthcare Hospital), Epsom and St Helier University Hospitals NHS Trust (Epsom Hospital), Manchester University NHS Foundation Trust (Manchester Royal Infirmary and Trafford Hospital), Medway Community Healthcare (Medway Community Hospital), Medway NHS Foundation Trust (Medway Maritime Hospital), Northumbria Healthcare NHS Foundation Trust (North Tyneside General Hospital and Wansbeck Hospital), Oxford University Hospitals NHS Foundation Trust (Horton Hospital and Nuffield Orthopaedic Centre), Royal United Hospitals Bath NHS Foundation Trust (Bath Royal United Hospital), Southern Health NHS Foundation Trust (Moorgreen Hospital) and The Royal Orthopaedic Hospital NHS Foundation Trust (The Royal Orthopaedic Hospital).

The CORKA trial team

Chief investigator: Professor Karen L Barker.

Co-investigators: Professor David Beard, Professor Gary Collins, Professor Avril Drummond, Professor Sarah Lamb, Professor Andrew Price, Dr Helen Campbell, Dr Francine Toye, Professor Martin Underwood and Dr Ly-Mee Yu.

Trial manager: Nicola Kenealy.

Trial statisticians: Dr Ruth Knight and Susan Dutton.

Health economists: Dr Jose Leal.

Qualitative: Dr Francine Toye.

Site principal investigators and research clinicians

Ashford Hospital: Leon Palmer-Wilson and Ana Glennon.

Countess of Chester Hospital: Helen Wilson.

Dorset County Hospital: Christian Brookes and Denise Hill.

Dorset Healthcare Hospital: Susan Dowdle and Hazel Burt.

Epsom Hospital: Jane Harrison.

Manchester Royal Infirmary: Sarah Adcock and Justine Theaker.

Medway Community Hospital: Gladys Nadar Arulmani.

Medway Maritime Hospital: Sunil Jain.

North Tyneside General Hospital: Mike Reed.

Oxford University Hospitals: Jonathan Room.

Royal United Hospitals Bath: Genevieve Simpson and Gemma Knight.

Moorgreen Hospital: Tricia Monroe.

Royal Orthopaedic Hospital Birmingham: Gareth Stephens and Sarah Rich.

Trial Steering Committee

Professor Anne Forster (Chairperson), Dr Derek Kyte, Dr Mark Kelson, Dr Mindy Cairns, Ms Rachel Dalton, Professor Jenny Butler, Professor Martin Underwood and Professor Avril Drummond.

Data Monitoring Committee

Dr Karen Smith (Chairperson), Professor Ruth Pickering and Dr Toby Smith.

Other acknowledgements

Dr David Smith and Tamsin Hughes for covering maternity leave of trial manager. David Kerr for trial data entry.

Cathy Jenkins, Erin Hannink and Martha Batting from Physiotherapy Research Unit, Nuffield Orthopaedic Centre, Oxford.

Dr Jennifer de Beyer of the Centre for Statistics in Medicine, University of Oxford, Oxford, for English-language editing.

Contributions of authors

Karen L Barker (https://orcid.org/0000-0001-9363-0383) (Professor of Physiotherapy) was the chief investigator, and led the funding application, trial conception and design, development of interventions, supervision, and writing and reviewing of the report.

Jon Room (https://orcid.org/0000-0002-1257-834X) (Research Physiotherapist) was the principal investigator, was involved in the development of the intervention, and writing and reviewing the report.

Ruth Knight (https://orcid.org/0000-0001-6810-2845) (Statistician) was involved in conducting the statistical analysis, and writing and reviewing the report.

Susan J Dutton (https://orcid.org/0000-0003-4573-5257) (Statistician) was involved in designing and conducting the statistical analysis, and writing and reviewing the report.

Fran Toye (https://orcid.org/0000-0002-8144-6519) (Qualitative Researcher) was the qualitative lead and was involved in conducting interviews and analysis, and writing Chapter 6 for the report.

Jose Leal (https://orcid.org/0000-0001-7870-6730) (Health Economist) was involved in designing and conducting the economic analysis, supervision, and writing and reviewing the report.

Seamus Kent (https://orcid.org/0000-0001-7298-3163) (Health Economist) was involved in conducting the economic analysis, and writing and reviewing the report.

Nicola Kenealy (https://orcid.org/0000-0002-3113-0305) (Trial Manager) was involved in trial management and writing and reviewing the report.

Michael M Schussel (https://orcid.org/0000-0002-1711-9310) (Statistician) was involved in screening tool development and writing for the report.

Gary Collins (https://orcid.org/0000-0002-2772-2316) (Co-applicant) supervised screening tool development.

David J Beard (https://orcid.org/0000-0001-7884-6389) (Co-applicant) provided clinical trials expertise and contributed to the report.

Andrew Price (https://orcid.org/0000-0002-4258-5866) (Co-applicant) provided surgical expertise and contributed to the report.

Martin Underwood (https://orcid.org/0000-0002-0309-1708) (Co-applicant) provided primary care expertise and contributed to the report.

Avril Drummond (https://orcid.org/0000-0003-1220-8354) (Co-applicant) provided occupational therapy expertise and contributed to the report.

Elaine Cook (https://orcid.org/0000-0001-5038-6034) (Co-applicant) was the patient and public involvement lead.

Sarah E Lamb (https://orcid.org/0000-0003-4349-7195) (Co-applicant) provided clinical trials expertise, was a key member of the trial management group, and was involved in writing and reviewing the report.

Publications

Barker KL, Beard D, Price A, Toye F, Underwood M, Drummond A, et al. COmmunity-based Rehabilitation after Knee Arthroplasty (CORKA): study protocol for a randomised controlled trial. Trials 2016;13:17.

Vadher K, Knight R, Barker KL, Dutton SJ. COmmunity-based Rehabilitation after Knee Arthroplasty (CORKA): statistical analysis plan for a randomised controlled trial. Trials 2018;19:19.

Barker KL, Batting M, Maia Schlussel M, Newman M. The reliability and validity of the Figure of 8 Walk test in older people with knee replacement: does the setting have an impact? Physiotherapy 2019;105:1.

Room J, Batting M, Barker KL. Development of a functional rehabilitation intervention for post knee arthroplasty patients: Community based Rehabilitation post Knee Arthroplasty (CORKA) trial. Physiotherapy 2020;106:52–64.

Data-sharing statement

All available data can be obtained by contacting the corresponding author.

Patient data

This work uses data provided by patients and collected by the NHS as part of their care and support. Using patient data is vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety, and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it’s important that there are safeguards to make sure that it is stored and used responsibly. Everyone should be able to find out about how patient data are used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health and Social Care.