Do pre-operative therapeutic interventions affect outcome in people undergoing hip and knee joint replacement? A systematic analysis of systematic reviews

Abstract Background THR and TKR patients represent one of the largest groups of surgical patients globally, yet we do not know how to optimise pre-operative care to improve post-operative outcomes. Objective To clarify the effect of pre-operative prehabilitation interventions such as exercise, neuromuscular stimulation, and psychological therapies on outcomes for hip (THR) and knee (TKR) replacement patients. Methods We used PRISMA guidelines and guidelines by Smith and colleagues on conducting reviews of reviews. Searches were conducted on Medline, the Cochrane Database of Systematic Reviews, and the Database of Abstracts of Reviews of Effects (DARE). Additional hand searches were also conducted. Articles were selected based on inclusion criteria and we report meta-analyses of data by outcome measures. Major Findings 6848 articles were screened and the full text of 33 reviews were obtained. Twenty systematic reviews were included, containing 67 unique randomised controlled trials. In 70% of the reviews (14/20), intervention fidelity was not reported. The components of prehabilitation that were tested were: exercise, education, nutrition, acupuncture and neuromuscular stimulation. Exercise alone did not affect functional outcome for TKR patients, but it did affect activity levels prior to THR, and pain prior to THR and TKR. Exercise alone may reduce Length of Stay (by between 0.8 and 4 days). Education when combined with exercise can reduce Length of Stay for TKR patients. Relaxation did not affect function or length of stay but gave a modest reduction of pain. Conclusions Providing education alongside exercise as a pre-operative intervention may reduce Length of Stay. One small RCT combined all three elements of exercise, education and dietary advice and there is no robust evidence to determine whether combining these elements can influence functional outcome.

Osteoarthritis (OA) is a leading cause of global disability with an age standardised point prevalence of 3754.2 per 100,000 [1].The management of hip and knee OA has been revolutionised by the advent of joint replacement (arthroplasty), which improves pain and quality of life [2].Arthroplasty is one of the most common surgical procedures in the UK, with 95,677 hip and 103,617 knee primary procedures carried out in England, Wales, and Northern Ireland in 2019 [3].This number reduced to 54,858 hip and 50,904 knee procedures in 2020 representing a massive under-provision [4].In 2020, waiting lists for elective orthopaedic surgery became 3 times longer than they were in 2019 largely due to the COVID pandemic [5].The situation has improved with 84,998 THR's and 77,830 TKRs carried out in 2021 [6] but the 2021 rates still represent a significant under provision and patients are still waiting longer [7].This heightens the need to improve outcomes and fully explain what high quality care should look like for orthopaedic patients who are waiting a long time for surgery so that we can maintain well-functioning health systems.
Prehabilitation refers to a series of structured interventions delivered before an operation which are designed to optimise patients' holistic fitness prior to surgery with the aim of improving outcomes [8].First studied and introduced in colorectal surgery [8], interest has now turned to evaluating prehabilitative interventions for patients undergoing elective orthopaedic procedures such as total knee replacements (TKRs) and total hip arthroplasties (THRs) [9].Within the context of orthopaedic surgery these are often exercise programmes, weightloss programmes, and psychological counselling, though numerous interventions have been studied [10][11][12].
Whilst individual studies have shown how exercise or physiotherapy programmes may be associated with lower post-operative pain, and a reduced length of hospital stay [13][14][15][16], a dose-response relationship has not always been conclusive [17].Educational and psychological interventions have also been explored with papers reporting efficacy in outcomes for depression, quality of life, and function after elective TKR and THR surgeries [17][18][19].
Providing a range of prehabilitation interventions prior to elective surgery may improve outcomes due to better management of patient expectations [19,20] (via education and counselling) and improved post-operative function, pain and muscle mass (via exercise.)[18] Lower body-mass index (BMI) has also been shown to help outcomes from TKR and THR surgery [21,22].
The National Institute for Health and Care Excellence (NICE) guidelines on joint replacements for hip, knee and shoulder [23] also emphasised the value of pre-operative care, However, NICE [23] has prioritised establishing the clinical and cost effectiveness of prehabilitation for arthroplasty patients, before publishing full guidance on what should be included in the orthopaedic prehabilitation intervention.Given the number of interventions studied, numerous systematic reviews have been published aiming to evaluate the effectiveness of prehabilitative interventions on surgical outcomes.Often these have focussed on one set of interventions such as exercise-alone and have rarely sought to evaluate multiple types of interventions.Smith et al. [24] have published on the need to provide robust evidence from these systematic reviews using new methodologies such as a review of reviews.
The aim of this review is to collate evidence from systematic reviews studying prehabilitation to summarise the effects, optimal mode of delivery, and clinical outcomes prior to hip and knee arthroplasty.

Methods
We use guidance published by Smith et al. and the PRISMA statement, which whilst pertaining to synthesis of individual trials, can also be used to categorise sub-sections of a review of reviews [24,25].
This review was registered with PROSPERO (CRD42019120886) at the University of York [26].

Inclusion criteria
Systematic reviews that assess any aspect of prehabilitation (education, exercise, nutrition, relaxation and/or psychological interventions) for people undergoing hip or knee arthroplasty were eligible.

Exclusion criteria
Non-English Language systematic reviews.

Search strategy
Our search strategy (Appendix 1) was designed by two reviewers, AR (an orthopaedic surgeon) and ELS (an orthopaedic physiotherapist).MeSH terms were used where applicable.The terms 'prehabilitation' (search strategy adapted from Perry et al. [27]); 'joint replacement' (adapted from Jordan et al. [28]) and 'systematic reviews' (adapted from White et al. [29]) were combined using Boolean operators following initial scoping searches.
Searches were conducted on 5-2-2019 (after an initial round of pilot searches) using Medline, the Cochrane Database of Systematic Reviews, the Database of Abstracts of Reviews of Effects (DARE) and the related articles function of PubMed from inception of databases.Hand searches from included papers and Google Scholar were conducted to maximise capture of literature.Searches were repeated on 15-4-2020, which identified Vasta et al. [30] and on 25-8-2021, which identified Dennis et al. [31].

Article selection and data extraction
Titles and abstracts were screened using Rayyan (Rayyan, USA) with two authors selecting papers based on the inclusion criteria.Deadlock was resolved through discussion with a senior author.A single author extracted data using Excel (Microsoft, USA).A second reviewer verified data extraction with disagreements resolved with the same senior author.
Interventions were grouped to facilitate reporting of outcomes.Two reviews (n ¼ 2) [32,33], included non-orthopaedic populations, which were not included in the analysis.Given that this paper aimed to synthesise evidence from papers with different study designs and inclusion criteria the I 2 statistic to determine heterogeneity was not used, and metaanalysis could not be employed.

Quality measurement
The AMSTAR 2 (Measurement Tool to Assess Systematic Reviews) [34] score was used to evaluate quality of each aspect of the review process as has been recommended when conducting a review of reviews.Risk of bias of each RCT was not assessed.

Results
The PRISMA flow chart is given in Figure 1.After removal of duplicates, we identified 6,848 titles and abstracts.Using inclusion and exclusion criteria this was reduced to 33 full text articles.Of these, 20 were included in the final analysis.We excluded 13 papers: not the right intervention or population (n ¼ 7); duplicated paper (n ¼ 1); not a systematic review (n ¼ 4); and article could not be obtained (n ¼ 1).The included papers were written in the United Kingdom (n ¼ 6), Australia (n ¼ 4), United States (n ¼ 4), Denmark (n ¼ 2), China (n ¼ 1), Canada (n ¼ 1), Italy (n ¼ 1) and the Netherland's (n ¼ 1).

Design of individual studies within the included reviews
The 20 systematic reviews included 67 unique clinical trials including 64 RCTs and 3 non-randomised showing the number of studies initially identified through database searches as 6,845 with duplicates removed.A further three studies were identified using hand searches leaving a total of 6,848 studies to be screened.33 articles remained after screening and 20 studies were selected for inclusion in the research.
A total of 5344 participants were included in the unique trials and studies reported in the selected twenty systematic reviews (range 82-3640 participants).Given that the 67 unique clinical trials, and seven studies of other designs were often duplicated in the included systematic reviews, tallying up participants from the reviews would have grossly overestimated the number of participants.Likewise, average age reported varied between from 48.0 years old to 83.0 years old with females reported between 12.2% and 66.0% in the analysed papers.

Prehabilitation interventions
Exercise and education were the two most commonly used intervention components with, 17 of the 20 reviews describing pre-operative exercise, whilst 13 reported on education.Other studies reported using acupuncture (2), neuro-muscular stimulation (NMS) (6), nutrition (3), and relaxation (1).Some of the reviews collated evidence from papers that used dual or tri-modality interventions with the most common combination being exercise and education.
In 70% of the reviews (14/20), compliance was not reported.None of the reviews described intervention 'monitoring' to ensure intervention delivery as planned.All 74 papers described the components of each intervention (Table 1).

Outcome measures
The systematic reviews reported the following outcomes:

Outcomes
Results are reported by intervention type.Ten reviews meta-analysed their data (Table 2).Significant and    .Meta-analysis could not be extracted from Baker et al. [46] but they identified no significant difference after pooling data.One review revealed no significant difference WOMAC stiffness sub-section (I2 ¼ 59%, MD À 0.26, 95% CI À 0.65 to 0.13, p > 0.05 -5 trials, 357 participants) [14].Three reviews show non-significant but slightly improved scores after prehabilitation involving exercise alone (a mix of home exercise and Physiotherapy supervised), at six and eight weeks [30,43], with larger differences favouring the intervention group at 12 weeks [17].One review reported significant within-group differences in three of their eight included studies [35].
Three non-meta-analysed reviews [30], identified no significant differences between the intervention and control groups.Two of these reported a nonsignificant weak effect of prehabilitation on function as scored by WOMAC between three weeks and six months post-operatively [30].
One review identified a statistically significant increase in the vitality component of the SF-36 for exercise combined with education, but no statistically significant difference in SF 36 for education only post-operatively [28].Two reviews which give non-significant results, demonstrate scores were consistently slightly improved in the prehabilitation group at 6 weeks post-operatively [17,30] .

Balance -TKR and THR.
Three reviews [17,45,46] referenced the same RCT by Gst€ oettner et al. with 18 participants [12] During the first 12 post-operative weeks' patients receiving proprioceptive training had improved standing balance.A repeated measures ANOVAs test tested the change between the three time points (baseline, after training, after TKA) achieving a P value of 0.045, demonstrating a change from baseline to after TKA.
McDonald [38] meta-analysed two RCT's combining pre-operative education with exercise for TKR patients reporting a reduced LOS by 1.86 days.

Education
Three reviews [38,41,42] focussed solely on education and conclude that pre-operative education may not produce post-operative patient benefit for THR and TKR patients, except when combined with exercise, which yielded a LOS reduction for TKR patients of 1.86 days (two trials, 183 participants, MD À 1.86; 95% CI À 3.40 to -0.32 [38]. Wallis et al. [37] examined the impact of combining exercise and education on pain.Two RCTs combining exercise and education (N ¼ 68) did not show any difference for pain before THR.One RCT showed improved pain at 12 weeks post operative (THR) with combined exercise and education, however the education was provided post-operatively (Ferrara et al. cited in Wallis).

Neuro-muscular stimulation
A RCT by Walls et al. [52] was analysed in five reviews [35], and one review [40] included the paper by Petterson et al. [53] Walls et al. [52] (with a population of nine TKR patients) found that while strength did not improve, stand-to-sit and stair climbing tests did at 12 weeks post operatively.

Acupuncture
Two reviews reported on Accupuncture [37,44], used pre-operatively.None of the studies demonstrated any significant effect on function, however one study found that the acupuncture group had significantly reduced pain at three months postoperative [54].

Nutrition
Six reviews reported the efficacy of nutrition and all referenced the same paper by Crowe and Henderson [51] which was the only RCT to provide nutritional support and dietician counselling to TKR patients.Whilst pre-operative anxiety, number of days to independently get out of bed and LOS were significantly reduced as reported in the reviews, this intervention was delivered in combination with exercise and hence the role of dietary advice in isolation cannot be assessed.Additionally, in this study the control group had significantly poorer function prior to receiving the intervention [38].

Relaxation
Four reviews included relaxation, which described two RCTs.Daltroy et al. [55] examined the effect of relaxation on THR and TKR patients and found it to be non-significant in influencing post-operative pain and anxiety.Patients received training in Benson's Relaxation Response and an 18-minute bedside audiotape the day before surgery.They were instructed on how to use relaxation to lessen anxiety and asked to practice before surgery.Berge et al. [56] taught relaxation for THR patients as part of an exercise program to improve the quality of rest, sleep and activity.This achieved a modest reduction in pain, and improved sleep, but no improvement in function.

Quality of the evidence and AMSTAR 2 scores.
Follow up in prehabilitation studies is generally inadequate [38], and we have validated this conclusion, finding that only a handful of studies within our 20 reviews followed up data collection for a year or more, with most limiting data collection to three to six months post-operative.Four reviews were high quality [33,38,39,48], eight were moderate [30,36,37,40,41,44,46,47], three were low [17,28,31] and four were of critically low quality [32,35,42,45] (Table 1).The most impactful omission was inadequate evaluation of bias, particularly on participant selection and blinding.Only one review was registered prior to completion.

Discussion
The main finding from this study is that a combination of interventions shortened LOS (for TKR patients).When the majority of studies meta-analysed evaluated exercise alone, no significant effect on LOS was observed Wang et al. [48].Similarly, meta-analyses of education when provided alonedoes not improve LOS [37].Yet, when meta-analyses included combinations of exercise and education, or exercise and acupuncture, better results were seen.Similarly, non-meta-analysed reports of individual studies combining multiple pre-operative interventions (Huang [50] and Crowe and Henderson [51] also show improvements in LOS. Only one (small) trial has combined three interventions (exercise, education and dietary counselling) [51], and more high-quality trials which consistently implement multi-modal prehabilitation are needed to identify whether dual or tri-modal prehabilitation should be provided for THR and TKR patients.Combining pre-operative interventions to safely promote earlier discharge and shorten LOS, could be vital at a time when the COVID-19 pandemic still places a magnifying glass on the core challenges on healthcare.
A second major finding is that exercise improved pre-operative activity levels (for THR patients) but not post-operative outcomes (including self-reported function) [Wallis et al.].Yet, this is contradicted by Chen et al. [14], with no improvement in pain postoperatively, after receiving an exercise programme before surgery.Exercise prior to THR or TKR surgery did not affect strength at all.By adding education to exercise, THR patients have improved pre-operative WOMAC and SF-36 scores.
Given the large number of papers we reviewed, it is surprising that dose and delivery method were poorly reported.Poor intervention reporting is not uncommon in surgery [57], and in other specialties [58] but reporting checklists can be used to improve this [59].Moving forwards, researchers should report the dose and delivery of prehabilitation interventions more fully to aid adoption of effective interventions into practice, and to ensure that studies can be effectively synthesised in systematic reviews.
Another recently published review of systematic reviews by Almeida et al. [9] who included ten systematic reviews on prehabilitation, is also available to clinicians.However, unlike our review, which is the first to offer a comprehensive review of dual and trimodality prehabilitation interventions, Almeida et al.
[9] only review the provision of exercise.In current times patients waiting for surgery tend to be less fit, with less exposure to weight bearing exercise and progression of symptoms.Therefore, although there is insufficient evidence to support each element of a multi-modal package, service providers advocate a multi-modal approach as it allows each intervention component to synergistically prepare patients for surgery [60].Some components seek to improve pain (acupuncture, relaxation), whilst others aim to increase muscle strength to facilitate a quicker return to function (exercise, NMS) or to improve knowledge/reduce anxiety to improve length of stay (education).
Socio-economic factors such as social status/deprivation, education and income may also affect outcome, yet none of the 20 reviews included them.It is perhaps surprising that frailty was only mentioned in one of the 20 systematic reviews we included -Milder et al. [32] This is a significant finding because arthroplasty is most commonly performed in patients who are older than 65 and identifying and adapting surgical care pathways in the light of patients' frailty has recently been emphasised by the new Centre for Perioperative Care guidelines [61].Finally, this paper only investigated the intervention stage of prehabilitation, the other three stages (screening, assessment, and monitoring) are not discussed here.
Due to the heterogeneity of intervention types, there are seemingly endless variables of who delivers it, duration, intensity and format.Therefore, future prehabilitation exercise trials should include a range of components, stretching, strengthening, range of motion and endurance exercises, (delivered in a home exercise plan, in a group, and face-to-face).Combining education with exercise has been shown to be beneficial when education was delivered faceto-face and supported with written materials [28,38].Specific modifications shown to give benefit, include adjusting educational material to suit learning styles [62], and tailoring elements to suit individual needs, providing unique resources for each patient.
Further work is required to evaluate the individual value of prehabilitation components -especially nutritional and psychological support which has been under researched in THR and TKR populations.While our question 'does prehabilitation improve outcomes in specific populations of THR and TKR patients?' has been answered here, with 'no, not always' -we argue that this conclusion should be interpreted with caution.Until a full-scale trial is conducted, which is purposefully designed to overcome design problems, we cannot know what the true impact of prehabilitation on orthopaedic populations might be.
This review has limitations.We used single data extraction which has been deemed inferior to double extraction (two reviewers independently extracting all the data).Yet, single data extraction allowed for pragmatism and for us to meet time contraints [63].Also, our data extraction sheet included quantitative data (such as number of education interventions), which is less likely to result in error than extracting more complex qualitative data [63].

Conclusion
Pre-operative therapeutic interventions can affect recovery after hip and knee arthroplasty, but the evidence is weak.We identified six intervention types: exercise, education, acupuncture, nutrition, relaxation and neuromuscular stimulation.Preoperative education may not provide post-operative benefit for THR and TKR patients except to reduce LOS [37,38].Exercise shortens LOS (between 0.8 and 4 days) and post-operative balance for TKR patients and pre-surgery functional status (timed chair rise, six-metre walk test, and timed up and go test) for THR patients.
Further work is required to evaluate the value of pre-operative nutritional and psychological support, acupuncture and neuro-muscular stimulation, and the effect of combining some or all of the six intervention types analysed in this paper as results were inconclusive or marginal.Acupuncture and relaxation therapies may improve pain whereas NMS and nutrition may improve function at 12-months though results are more difficult to interpret based on reviews due to multi-modality interventions.
This paper identifies commonly used outcome measures and recommends the development of a core outcome set to support future meta-analysis in this exciting and developing field.

Figure 1 .
Figure 1.PRISMA Flow diagram.Evaluating the effect of pre-habilitation on outcome in orthopaedic surgery.PRISMA Flow diagramshowing the number of studies initially identified through database searches as 6,845 with duplicates removed.A further three studies were identified using hand searches leaving a total of 6,848 studies to be screened.33 articles remained after screening and 20 studies were selected for inclusion in the research.
This is a table showing the included 20 studies being screened via AMSTAR 2 tool for quality.Four studies were deemed of high quality, eight of moderate quality, four of low quality, and four of critically low quality.
RCTs, Randomised Control Trials (including quasi-randomised trials); NRS þ RTs, Non-Randomised Studies and Randomised Trials; Y, yes; N, No; PY, Partial Yes; x, No meta-analysis performed; � , Could only be scored if a meta-analysis was performed.

Table 2 .
Continued.-significant results for each outcome are described when meta-analysis was not used.Economic outcomes were not reported, however hospital length of stay was often given as a marker of service utilisation.Where abbreviations are used MD ¼ Mean differences, SMD ¼ Standardised mean difference, WMD ¼ Weighted Mean Difference.
24)ervention and standard care in terms of longterm pain as assessed on the KOOS (MD À 2.55, 95%CI À 6.35 to 1.24)This is a long-table with multiple domains showing the data that has been extracted from each of the 20 systematic reviews.Where systematic reviews analysed underlying Randomised Control Trials (RCTs), the number of RCTs has been included too.Further the domain of prehabilitation each paper has analysed, whether that be education, exercise, acupuncture or neuromuscular stimulation has been included in the table.LEGEND: ROM Range of motion, MA meta-analysis, THR Total Hip Replacement, TKR Total Knee Replacement, ortho orthopaedics, WOMAC The Western Ontario and McMaster Universities Arthritis Index, HSS Hospital for Special Surgery Survey, Grey rows indicate non meta-analysed results, LOS Length of Stay.non95% CI: 26.38, 3.33; 2 trials -Simmons et al. [45]exercise and NMS); (SMD À 0.15 (95% CI, 0.41 to 0.11; I2, 0.0%, P for heterogeneity ¼0.478, 6 studies, 230 patients -Hoogeboom et al. [39]-exercise alone); (SMD À 0.06, 95% CI [À 0.39, 0.26] 2 trials, 143 participants -Cabilan et al. [33] -exercise alone) Petterson et al. [53] found significant improvement in voluntary muscle activation up to 52 weeks post operative with NMS delivered at prehabilitation.Walls et al. provided NMS over 37 sessions and Petterson et al. [53] gave 17 sessions combined with home-based exercise.