THE PLACE OF ARTHROSCOPY IN DEGENERATIVE KNEE PAIN WITHOUT TRUE LOCKING IN ADULTS

Transcription

1 1 EVIDENCE SUMMARY REPORT THE PLACE OF ARTHROSCOPY IN DEGENERATIVE KNEE PAIN WITHOUT TRUE LOCKING IN ADULTS Question to be addressed Is there evidence that arthroscopic investigation and treatment of the knee is more clinically and cost effective than conservative management for adults with generalised knee pain? Reason for Review Solihull, Birmingham CrossCity and Birmingham South Central CCGs (BSOL) requested a rapid evidence review (and analysis of recent recorded activity) of the clinical and cost effectiveness of knee arthroscopy for degenerative knee pain. The detailed scope of the evidence review was confirmed by BSOL CCG commissioners in July Options for commissioners to consider: 1. To commission arthroscopic lavage and debridement, with or without partial meniscectomy, only for people with both persistent knee pain and a clear history of persistent mechanical locking. 2. Arthroscopic lavage and debridement, with or without partial meniscectomy, will not usually be commissioned for patients with degenerative knee disease including radiographic and other symptoms of osteoarthritis, meniscus tears and mechanical symptoms. 3. To continue to commission arthroscopic lavage and debridement, with or without partial meniscectomy, at the discretion of clinicians. Summary Background The most common cause of generalised knee pain is osteoarthritis (OA). OA is the result of progressive degeneration of the cartilage of the joint surface. Meniscal tears and other structural changes including osteophytes, cartilage and bone marrow lesions are common characteristics of knee osteoarthritis. The condition is also known as degenerative knee disease. The relationship between degradation of the joint surfaces and knee osteoarthritis is unclear. Imaging abnormalities of the knee surfaces are common and are known to exist in pain-free knees as well as symptomatic patients. Radiographic signs of osteoarthritis can be accompanied by pain and stiffness, and can impair patients ability to perform activities of daily living and recreational activities. Conservative treatments aimed at reducing the symptoms include patient information, weight loss and physical therapy. NICE advises that pain relief medication, physiotherapy, arthrocentesis and intra-articular corticosteroid injections may also be beneficial if the pain is moderate to severe.

2 2 EVIDENCE SUMMARY REPORT Arthroscopic knee surgery is an established and common treatment option and may include arthroscopic lavage (also called arthroscopic washout ), arthroscopic debridement (in combination with lavage) and arthroscopic partial meniscectomy (APM) which may be performed singly or in combination with debridement and lavage. A large volume of evidence has led to NICE recommendations that arthroscopic lavage and debridement should not be used in knee OA. Arthroscopy procedures are still commonly performed with approximately 150,000 procedures being carried out in the UK each year. APM is the most common knee procedure performed in the UK (151.2 procedures per 100,000 population). Clinical effectiveness There is published evidence from one recent and well conducted SRMA (Brignardello- Petersen et al 2017) and a well conducted RCT which compared arthroscopic surgery with sham arthroscopic surgery showing that arthroscopic debridement with or without partial meniscectomy is not superior to conservative management. The SRMA found that there was a short term improvement in pain and function at 3 months but this benefit was not sustained at 12 and 24 months. o Pain: The mean difference (MD) in favour of APM was only 5.38 more points (95%CI 1.9 to 8.8) out of 100 and, after 1000 procedures, it was estimated that only a small number of additional patients (124) would gain at least 12 points minimally important improvement in pain at 3 months. o Function: The mean difference in favour of APM was 4.9 (95%CI 1.5 to 8.4) points. An additional 134 patients per 1000 procedures achieved the minimally important difference (MID) of 8 points for function as a result of surgery. o The short term benefit in pain and function did not translate to an improvement in QoL. The outcomes from the meta-analysis of low quality studies are reinforced by the findings from the FIDELITY trial which compares surgery with sham surgery. This corrects for the inherent bias and preferences of patients and carers who may have an initial preference for surgery (and consider non-surgical interventions to be inferior). o There was no difference between the APM and sham surgery at any time point (2,6,12,24 months) for any primary or secondary outcomes including pain or function and adverse events measured by any knee score. The baseline presence of mechanical symptoms or meniscal tear did not result in APM being more favourable. There was no difference between the APM and conservative management (or sham surgery) groups for: o Change in activity level o Development/progression of OA o Adverse events o General health Initial presentation with mechanical symptoms. Safety There is weak evidence from a recent well-conducted systematic review and meta-analysis by Brignardello-Petersen et al (2017) and from two large retrospective cohort studies that knee arthroscopy carries a low risk of adverse events. The most commonly reported adverse events are venous thromboembolism (VTE) and infection (occurring in 5 per 1000 procedures and 2 per 1000 procedures respectively). It is not clear that the short term benefits associated with APM outweigh the (small) risks.

3 3 EVIDENCE SUMMARY REPORT Cost effectiveness There is no reliable evidence which supports the use of arthroscopic knee surgery for people with degenerative knee disease. The most reliable study available concludes that the procedure is not cost effective despite including some indirect costs and only assessing costs effectiveness up to two years after surgery. The cost effectiveness of knee arthroscopy treatment is highly dependent upon the clinical effectiveness. Therefore, it is not possible for the procedure to be cost effective if it is only marginally clinically effective, especially in view of the evidence that it may, in a very small proportion of cases, be harmful. Activity and finance For the 27 months from April 2015 to June 2017 inclusive, there were 2,036 elective admissions for knee arthroscopy 1 for patients registered with a BSOL CCG GP. This activity excludes admissions which included arthroscopic lavage, debridement or partial meniscectomy as part of a more complex knee procedure such as ligament repair and for procedures or diagnoses out of scope of this review. The most commonly performed knee arthroscopy procedure was W822: endoscopic resection of semilunar cartilage NEC. This accounted for 1,617 (79%) of the 2,036 admissions and collectively cost the three CCGs within the BSOL area 4,496,357 (82% of the total cost of knee arthroscopy admissions) in just over two years. 1 Defined as arthroscopic lavage, arthroscopic debridement (in combination with lavage) and arthroscopic partial meniscectomy which may be performed singly or in combination with debridement and lavage.

4 4 EVIDENCE SUMMARY REPORT 1 Context 1.1 Introduction In older people, the most common cause of generalised knee pain is osteoarthritis (OA) which is the focus of this review. Osteoarthritis of the knee is the result of progressive degeneration of the cartilage of the joint surface. Meniscal tears and other structural changes including osteophytes, cartilage and bone marrow lesions are common characteristics of knee osteoarthritis. The condition is also known as degenerative knee disease. The relationship between degradation of the joint surfaces and knee osteoarthritis is unclear. Imaging abnormalities of the knee surfaces are common and are known to exist in pain-free knees in middle-aged and older people [1, 2] as well as symptomatic patients. Radiographic signs of osteoarthritis can be accompanied by pain and stiffness, and can impair patients ability to perform activities of daily living and recreational activities [3]. Osteoarthritis is a long-term condition and cannot be cured. A number of treatments are available to reduce the symptoms of osteoarthritis, the choice of which will depend on the severity of the osteoarthritis. Initial core treatments [4] recommended by NICE before referral for consideration for surgery include: ongoing verbal and written information about the condition and its management advice on physical activity and exercise for muscle strengthening and general fitness support to lose weight if the person is overweight or obese 2. Other conservative treatments include self-management, medications to relieve pain and inflammation, and physiotherapy. If there is a knee-joint effusion, fluid around the knee may be aspirated with a needle (arthrocentesis) to reduce pain and swelling. Intra-articular corticosteroid injections should be considered as an adjunct to core treatments for the relief of moderate to severe pain in people with osteoarthritis (NICE guideline). In a small number of cases, where the above treatments are inadequate or the damage to the joints is particularly severe, surgery may be carried out to repair, strengthen or replace a damaged joint. There are several treatments for patients with osteoarthritis of the knee. This review includes three commonly performed arthroscopic knee procedures: arthroscopic lavage (also called arthroscopic washout ), arthroscopic debridement (in combination with lavage) and arthroscopic partial meniscectomy (APM) which may be performed singly or in combination with debridement and lavage [5]. A well-publicised randomised controlled trial by Moseley et al in 2002 [6], combined with a Cochrane systematic review of the literature up to 2006 [7], led to NICE guidance which recommended that arthroscopic lavage and debridement should not be used in knee OA (see below). Despite the existence of high quality evidence and national guidance, a 2015 editorial [8] stated that arthroscopy procedures are still commonly performed with approximately 150,000 procedures being carried out in the UK each year. Researchers also report that although there was an 80% reduction in arthroscopic irrigations which decreased by 39.6 procedures per 2 Clinical guideline CG177 recommendations 1.2.5, 1.3.1, and 1.4.3

5 5 EVIDENCE SUMMARY REPORT 100,000 population between 2000 and 2012, this corresponded with an increase in the rate of arthroscopic meniscal resections (APM) by per 100,000 population or 230%. The largest absolute number of arthroscopic knee procedures performed between 2000 and 2012 was for meniscal resection (151.2 procedures per 100,000 population) [9]. 1.2 Existing national policies and guidance In August 2007, NICE published interventional procedure guidance on the effectiveness of arthroscopic knee washout, with or without debridement, for the treatment of osteoarthritis (IPG230) [10]. The guidance recommended that joint lavage, used alone, without debridement, for patients with knee osteoarthritis is not effective and should not be used. 1.1 Evidence on the safety and efficacy of arthroscopic knee washout with debridement for the treatment of osteoarthritis is adequate to support the use of this procedure provided that normal arrangements are in place for consent, audit and clinical governance. 1.2 Current evidence suggests that arthroscopic knee washout alone should not be used as a treatment for osteoarthritis because it cannot demonstrate clinically useful benefit in the short or long term. Since then, subsequent guidance from NICE has been issued. This included the clinical guideline (CG177), Osteoarthritis: care and management published in February 2014 which further refined the recommendation to exclude debridement as well as lavage unless there was mechanical locking Do not refer for arthroscopic lavage and debridement as part of treatment for osteoarthritis, unless the person has knee osteoarthritis with a clear history of mechanical locking (as opposed to morning joint stiffness, 'giving way' or X-ray evidence of loose bodies). [2008, amended 2014] This recommendation is reflected in the surgical options section of the NICE Pathway for the management of osteoarthritis [11] (Figure 1 below) and later as examples of NICE do not do case studies.

6 6 EVIDENCE SUMMARY REPORT Figure 1: NICE Pathway for the Management of Osteoarthritis Source: NICE Pathway for the Management of Osteoarthritis (11). Subsequent to the NICE guidelines, a European consensus guideline (ESSKA) 3 was published in February 2017 [32]. The main finding was that arthroscopic partial meniscectomy should not be proposed as a first line of treatment for degenerative meniscus lesions. Arthroscopic partial meniscectomy should only be considered after a proper standardised clinical and radiological evaluation and when the response to non-operative management has not been satisfactory. Magnetic resonance imaging of the knee is typically not indicated in the first-line work-up, but knee radiography should be used as an imaging tool to support a diagnosis of osteoarthritis or to detect certain rare pathologies, such as tumours or fractures of the knee. [32] The literature review which informed the ESSKA consensus guideline included papers published to May However, the process for reaching consensus guidelines does allow for departures from evidence findings as well as pragmatic consensus where evidence is absent or unreliable. 3 European Society for Sports traumatology, Knee Surgery and Arthroscopy

7 7 EVIDENCE SUMMARY REPORT The recent BMJ Clinical Practice Guideline is based on the 2017 systematic review by Brignardello-Petersen et al [19]. This review, with a search date of June 2017 includes the results of four more recently published studies in addition to the Brignardello-Petersen systematic review. 2 Epidemiology Radiographic OA of the hands, knee, and hip joints is frequent in European populations, especially in those over the age of 50 years. About 8.5 million people in the UK have OA, and this figure is likely to increase due to an ageing population [12]. OA becomes more common with advancing age. The Framingham Osteoarthritis Study showed that the prevalence of radiographic OA increases with age from 27% in people younger than 60 years to 44% in those older than 70 years [13]. Knee OA affects more than 10% of the population over 60 years old [14]. It is more common in women. The Framingham study showed that 2% of women with a mean age of 71 years develop radiographic knee OA every year, compared with 1.4% of men, and 1% of women develop symptomatic knee OA every year, compared with 0.7% of men. In a general practice UK study, the prevalence of currently recorded diagnosis of knee OA in patients over 45 years old was 1.1%, and the estimated prevalence of all those currently registered with the practice who had had knee OA diagnosed at some point was 5.5%. 3 The intervention Arthroscopic lavage, debridement and/or partial meniscectomy involve the removal of loose bodies or osteophytes in the knee [15] and are sometimes used to treat knee OA [10]. In the USA and in the UK, treatment of meniscal tears is the most common reason for knee arthroscopy Other indications for arthroscopic knee surgery which are outside of the scope of this review include cruciate ligament reconstruction, osteochondral lesions, synovectomy, and septic arthritis. Arthroscopic lavage or arthroscopic lavage and debridement with/or without partial meniscectomy are usually performed under general anaesthesia. Arthroscopy is accomplished with the use of two small incisions on either side of the patella at the anterior aspect of the knee. An arthroscope is inserted through one incision and used to view the image on a monitor, while the second incision is employed for instrumentation. Prior to the initiation of the procedure, diagnostic arthroscopy is performed. Meniscal tears and other pathology are identified. With use of an arthroscopic probe, the features of the meniscal tear are determined. Arthroscopic punches and shavers are used to debride torn portions back to a stable rim. Washout expels loose debris through the cannula. At the end of the procedure, the saline is drained out of the joint and the incisions are closed with stitches [16].

8 8 EVIDENCE SUMMARY REPORT 4 Findings We searched PubMed, Embase, Cochrane Library, TRIP and NICE Evidence on the 23 rd June 2017 using the search strategy detailed in section 8 below. For arthroscopic lavage only, we searched for studies published after the NICE IPG (published in 2006). A previous SPH evidence review on this topic (search date December 2016) identified a key systematic review and meta-analysis by Thorlund et al 2015 [17] which focused on arthroscopic debridement and/or partial meniscectomy.we therefore restricted the search for these interventions to 2016 onwards. The search was also limited to English language publications and we excluded conference papers, letters, commentary and editorials. 4.1 Evidence of effectiveness We found two directly relevant systematic reviews and meta-analyses (SRMA). The first of these was a Cochrane systematic review [18] published by Reichenbach et al (2010) and focused on the clinical effectiveness of knee joint arthroscopic lavage only. We found no new trials for arthroscopic lavage alone published since this systematic review. The second, by Brignardello-Petersen et al [19] was published in February 2017 and included subjects who had arthroscopic debridement and/or partial meniscectomy. The search date was 16 th August This review included 13 randomised controlled trials (RCTs) for effects of knee arthroscopy and 15 studies (3 RCTs, 12 observational studies) for complications of knee arthroscopy. It superseded other recent SRMAs including both Thorlund et al (2015) [17] and Van de Graaf et al (2016) [20]. In addition, we found outcomes from three RCTs reported in four more recent publications [21, 22, 23, 24] than the search dates of these SRMAs. The outcome periods reported ranged from six months to three years. In addition to the complications reported in the SRMA by Brignardello-Petersen (2017) [19], we found adverse event outcomes in two large retrospective US registry studies that investigated risk factors and perioperative outcomes in patients who had undergone arthroscopic knee procedures [25, 26]. Despite the established nature of both the condition and the intervention, we found only two relevant studies about the cost effectiveness of arthroscopic surgery for OA knee [27, 28]. The identified studies are all described in Table 1 below Clinical effectiveness of arthroscopic investigation and treatment of the knee For arthroscopic lavage alone (without debridement or partial meniscectomy), the Cochrane systematic review by Reichenbach et al in 2010[18], reported no significant improvement in pain or function at three months and at one year. This review included seven trials (n=567), all with

9 9 EVIDENCE SUMMARY REPORT methodological limitations (small, few in number, poor quality, heterogeneous in methodology and outcomes). The authors concluded that joint lavage does not result in a relevant benefit for patients with knee osteoarthritis in terms of pain relief or improvement of function. The systematic review and meta-analysis by Brignardello-Petersen et al (2017)[19] reviewed the clinical and cost effectiveness of arthroscopic debridement and/or partial meniscectomy compared to any conservative management strategy for patients with degenerative knee disease. This included patients with or without: Imaging evidence of osteoarthritis Meniscus tears Locking, clicking or other mechanical symptoms Acute or subacute onset of symptoms (but excluding patients with symptoms after major knee trauma with acute onset of joint swelling). The review included 1668 patients from 13 RCTs which followed up patients for between three and 24 months. The primary outcome measures were patient-reported pain, physical function and quality of life (QoL).The review also assessed harms reported in three of the RCTs and an additional 12 observational studies (more than 1.8 million patients overall). The study populations included a range of severity of OA and all arthroscopic procedures involving partial meniscectomy, debridement or both. This reflected the range of clinical practice currently followed. The key findings were that arthroscopic debridement and/or partial meniscectomy provides a small benefit to patients in reducing pain for up to three months. The mean difference (MD) in favour of APM was 5.38 more points (95%CI 1.9 to 8.8). The meta-analysis estimated that an additional 124 patients per 1000 procedures achieved the 12 point minimally important difference (MID 4 ) in pain reduction. This pain improvement was not sustained; there was no important difference between the groups at one to two years (APM vs conservative management: 22 vs 19 points (MD (95%CI to 6.43)). Similarly, arthroscopic surgery was associated with a small benefit in improving function in the short term (APM vs conservative management: 14 vs 9 points (MD 4.9 (95%CI 1.5 to 8.4)). An additional 134 patients per 1000 procedures achieved the minimally important difference (MID) of 8 points for function as a result of surgery. At one to two years, there was no longer an important difference between the groups for function (APM vs conservative management: 13 vs 10 points (MD 3.16 more (95%CI to 6.8)). Quality of life (QoL) was not reported in the majority of the RCTs included in the SRMA. As the QoL analysis was based on only one study (n=120) at three months and two studies (n= 269) at one year, the results are unreliable and should be considered with caution. At both three months and one year, there was no important difference (MID for EQ-5D index is 15 points) between arthroscopic surgery and conservative management for QoL: At three months: 14 vs 8 points (MD 6.0 (95%CI -1.5 to 13.5)) At one year: 12.4 vs 10.3 points (MD 2.12 (95%CI to 5.21)). 4 Minimally important difference (MID). This was based on a systematic review and case study by Devji et al (2017) [34] and addressed what level of individual change on a given scale is important to patients (minimally important difference). The study recommended that the minimally important difference for pain and for function was 12 points and 8 points respectively. 5 The mean difference (MD) measures the absolute difference between the mean values of the intervention (arthroscopy) versus control (conservative management) groups.

10 10 EVIDENCE SUMMARY REPORT For up to one year after knee arthroscopy surgery, there were more patients who had subsequent knee replacement surgery: 23 vs 12 per 1000 procedures (MD 11 per 1000 (95%CI 1.07 to -6)). This SRMA was well conducted, with individual studies assessed for bias and quality. However, it had some weaknesses, as there was heterogeneity between the included studies, bias and incomplete reporting (particularly of possible subgroups of interest such as those with mechanical locking). The interventions included debridement and partial meniscectomy. There were a number of permutations of interventions including: Arthroscopic partial meniscectomy (APM) alone Debridement with APM Debridement, APM and exercise The control groups of the RCTs also varied: Arthroscopic lavage Sham surgery Medical therapy Exercise (e.g. physiotherapy, supervised exercise, at home exercise; not necessarily optimal ) All of the RCTs included in the review were small. There was heterogeneity between these studies in that the interventions included APM, debridement or both, the controls varied and the follow-up time for the primary outcome ranged from three to 24 months. In five of the 13 RCTs, more than 50% patients had radiographic osteoarthritis as well as symptoms. However, evidence for pain outcomes is considered high quality; the risk of bias analysis found no statistically significant or clinically meaningful differences in effect size for pain between studies with adequate or inadequate blinding. For function, the evidence was of moderate quality only; the included trials were at a higher risk of bias and the measures of function were considered imprecise. The authors report that the presence of inadequate blinding was likely to overestimate the effect of arthroscopic surgery as there is a stronger placebo effect for surgical intervention over noninvasive procedures. Despite this, they found the benefit of arthroscopic meniscectomy (with or without debridement) was not statistically or clinically significant beyond three months. The overall evidence base for arthroscopy surgery (lavage and debridement/partial meniscectomy) is weak and based on thirteen small RCTs. The value of the meta-analysis is that it reviewed the best of the weak evidence available and attempted to correct for a variety of confounders and bias. A major strength of this review is that the authors converted different pain, function and QoL instruments 678 to the scale of a single index instrument (0-100) for each. This allowed for an 6 Lysholm Knee Scoring Scale (8 questions for pain and function: 0 to 100 marks with fewer marks for more severe symptoms). It includes a locking domain question where patients are asked to choose one out of the following 5 responses that best reflected the status of their knee: i) no locking or catching ii) catching sensations but no locking, iii) occasional locking, iv) frequent locking or v) locked at present. 7 Knee Injury and Osteoarthritis Outcome Score. KOOS consists of 5 subscales; Pain, other Symptoms, Function in daily living (ADL), Function in sport and recreation (Sport/Rec) and knee related Quality of life (QoL), each scoring 0 to 4

11 11 EVIDENCE SUMMARY REPORT estimate of the proportion of patients who would achieve a minimally important difference in outcome. The authors concluded that the review provided low quality evidence that the procedure provides very small benefits in pain and function in the short term over conservative therapy [19]. This systematic review is the most comprehensive and reliable source of evidence available. There were four publications reporting outcomes from three RCTs which were published after the SRMA by Brignardello-Petersen et al [21, 22, 23, 24]. The Finnish, multi-centre RCT (FIDELITY) by Sivhonen et al [21, 22] studied the clinical effectiveness of APM compared to sham surgery. All patients selected for inclusion had confirmed degenerative medial meniscus tear, with nearly half of the subjects in each group reporting some catching or locking symptoms, although none had a bucket handle meniscal tear which is usually associated with a true locked knee. At one year, there was no difference for mechanical (locking) symptoms between APM and sham surgery for either the whole cohort or specifically for those patients who initially reported locking, indicating that APM has no significant benefit over sham surgery to relieve knee catching or occasional locking [22]. At two years, there was no significant difference between APM vs placebo for the following knee outcome measures 9 (APM mean (95%CI) vs placebo (mean (95%CI), between group difference in improvement from baseline mean (95% CI)). WOMET 10 : 80.9 (95%CI 75.4 to 86.5) vs 86.1 (95% CI 80.5 to 91.8), -5.2 (95%CI to 2.7) Lysholm knee score: 82.2 (78.2 to 86.3) vs 86.5 (82.3 to 90.6), -4.3 (-10.0 to 1.5) Knee Pain after exercise 11 : 2.3 (1.5 to 3.1) vs 1.9 (1.1 to 2.7), -0.4 (-1.5 to 0.7) There was no significant difference between the APM and placebo surgery group for any of the secondary outcomes: Satisfied patients Improved patients Treatment group unblinding Reoperations including arthroscopy, tibial osteotomy or total knee arthroplasty Return to normal activities Serious adverse events Mechanical symptoms Meniscal tests 8 The Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) is a composite knee assessment score. It assesses pain, stiffness, and physical function in patients with hip and / or knee osteoarthritis. The WOMAC consists of 24 items divided into 3 subscales: Pain (5 items), Stiffness (2 items) and Physical Function (17 items). 9 Values are adjusted with the baseline score, study site, age, sex and the absence or presence of minor degenerative changes on a radiograph (Kellgren-Lawrence grade 0 or 1, respectively). 10 Western Ontario Meniscal Evaluation Tool (WOMET) is a meniscus specific health related quality of life instrument validated especially for patients with degenerative meniscal tear. It includes 16 items addressing 3 domains: 9 items addressing physical symptoms; 4 items addressing disabilities with regard to sports, recreation, work and lifestyle, and 3 items addressing emotions. WOMET scores are converted to a percentage of normal, and therefore range from with 0 indicating the most severe symptoms and 100 the absence of symptoms. 11 Knee pain after exercise (during the preceding week) was assessed on an 11 point numerical rating scale ranging from 0=no pain to 10=extreme pain

12 12 EVIDENCE SUMMARY REPORT The authors of this APM vs sham APM surgery trial attempted subgroup analysis of those patients who had been recruited to the study with mechanical symptoms or an unstable meniscus tear 12. Both subgroups had far fewer subjects in each arm. In patients who presented with mechanical symptoms, there was no difference between APM (n=32) and placebo surgery (n=37) groups for any of the primary outcomes: WOMET: p=0.113 Lysholm knee score: p=0.268 Knee pain after exercise: p=0.097 In patients who presented with an unstable meniscus tear, there was no difference between APM (n=34) and placebo surgery (n=39) groups for any of the primary outcomes (WOMET, Lysholm knee score and knee pain after exercise). Neither subgroup showed any difference for any secondary outcomes including the number and proportion of satisfied patients, improved patients, treatment group unblinding, returned to normal activities, mechanical symptoms, those with at least one positive meniscal test. A Swedish, single centre RCT by Gauffin et al (2017) [24] already included in the SRMA, published three year outcomes in From the initial 150 patients recruited, 75 were allocated to received knee arthroscopy (APM) and a three month exercise programme (PT), whilst the control group received exercise therapy alone. The authors reported that 25% of subjects crossed over to receive surgery (16 before the end of one1 year, and a further 3 at the 3 year analysis). The authors found that at three years, a greater proportion of patients who received APM achieve at least a 10 point improvement in the KOOS Pain score 13 (APM vs PT): 81% (50/62) vs 73% (41/56) compared to a proportion of patients allocated to exercise therapy alone, reporting more than a 10 point deterioration in pain (APM vs PT): 0% (0/62) vs 11% (6/56), p=0.03. There was also a significant difference between APM vs PT groups for symptoms and for quality of life: KOOS symptoms: mean 8.3 (95% CI 0.9 to 15.8), p=0.029 KOOS EQ-5D Index: mean 0.13 (95% CI 0.03 to 0.23), p=0.016 However, the improvement in proportion of people achieving a greater than 10 point KOOS pain score, and the between group improvement in mean score for symptoms and EQ-5D, is not consistent with or supported by significant improvement for: KOOS pain: p=0.068 KOOS ADL; p=0.08 KOOS Sports: p=0.406 KOOS QoL: p=0.062 EQ-5D EQ-VAS: p=0.353 It is possible that a subgroup of people achieve greater pain improvement. The authors investigated the outcomes for patients who had an improvement in KOOS pain score (APM n=59 12 Meniscus tears with longitudinal tear pattern, bucket handle tear or flap were determined as unstable 13 >10pt improvement in KOOS pain score (considered a clinically important improvement

13 13 EVIDENCE SUMMARY REPORT vs PT n=52). The between group difference at one year (APM vs PT: 10.9 points, 95%CI 3.3 to 18.4, p=0.005) 14 was not sustained at three years (p=0.137). The authors also reported between group difference in people who had an improved pain score for older patients aged (p=0.02), and for those who presented without mechanical symptoms (p=0.017). The outcomes for this study should be treated with caution. The study was single centre, it was not possible to blind the patients for the treatment, and there was a significant proportion of crossover. Subgroup analysis is inherently unreliable and subject to bias. One additional RCT (7 centres, USA) was set up to examine the incidence and reasons for crossover [23]. This study recruited patients for APM and PT (n=164) or PT alone (n=177). The study had difficulty recruiting patients with only 26% of eligible patients being randomised. This is because PT or non-surgical interventions are perceived as being an inferior treatment option by the patient, even before experiencing it. Consistent with this, Katz et al (2016) [23] reported that 48 subjects (27%) had crossed over to the surgery arm with 140 days. At six months, a similar proportion of patients who had APM (or who had crossed over to APM (82% and 81% respectively)) achieved a greater than 10 point improvement in KOOS pain score (RR=0.95 (95%CI 0.64 to 1.41)) compared to PT (73%). There was no difference in the extent of pain improvement between the APM and crossover to APM groups (p=0.41). They reported that factors associated with crossover to APM were duration of symptoms of less than one year (risk ratio 1.74 (95% CI 0.98 to 3.08)) and separately a baseline WOMAC pain score greater than 40 points (risk ratio 1.99 (95% CI 1.00 to 3.93)). Factors which were found not to be associated with crossover to APM included age, sex, body mass index (BMI), meniscal symptoms (aggregated or separated as intermittent locking or catching), medical comorbidities, mental health and physical features Safety The systematic review and meta-analysis by Brignardello-Petersen et al (2017)[19] also reported the possible harms caused by arthroscopic surgery including any or all of debridement and/or partial meniscectomy compared to conservative treatment (exercise therapy, injections, drugs, sham surgery, no treatment). Three months after surgery, the number of adverse events per 1000 procedures was low (arthroscopic surgery vs conservative treatment): Mortality: no deaths in either group per 1000, less than 1 more death per 1000 procedures (95%CI 0 to 1) Venous thromboembolism: 5 vs 0 per 1000, 5 more events per 1000 (95%CI 2 to 10) Infection: 2 vs 0 per 1000, 2 more events per 1000 (95%CI 1 to 4) Nerve damage:0 vs 0 per 1000, less than 1 more event per 1000 (95%CI 0 to1) 14 This is not consistent with the KOOS pain scores at 1 year based on the larger cohort (APM n=70 vs PT n=60): 84 (81-88) vs 78 (73-83), p=0.029

14 14 EVIDENCE SUMMARY REPORT The meta-analysis of complications was based on three of the RCTs already included in the systematic review of effectiveness, one prospective study and eleven large retrospective registry studies (the largest five of these comprised over 1,500,000 subjects collectively). There was considerable heterogeneity between these studies in study size (which ranged from 70 to 432,038 subjects), design, patient characteristics (mean age ranged from 42 to 62.4 years) and comparator treatments. In addition, some adverse events were estimated on fewer than the 15 studies reviewed overall. Although the heterogeneity indicates that the estimates of the degree of harm should be treated with caution, the conclusion that arthroscopic meniscectomy with or without debridement is not risk free is reasonable. In addition, we found two large retrospective cohort studies, both from the USA [25, 26]. These are summarised in Table 1. The largest cohort study included 330,714 patients who had one of 13 arthroscopic knee procedures, the most common of which was APM (66.5%, n=219,964) between 2007 and 2011 [25]. Patients who had APM had the lowest incidence of all the 13 arthroscopic knee procedures for both manipulation under anaesthetic (0.06% at 3 months, 0.11% at 6 months) and lysis of adhesions (0.04% at 3 months, 0.06% at 6 months). There is weak evidence from a smaller cohort study (n=15,167) of both arthroscopic shoulder and knee procedures that the 30-day readmission rate for knee procedures was low (0.92%) and, of these, arthroscopic debridement represented only 1.56%. No readmission rate was reported specifically for APM [26]. The quality of the evidence for harms is low. The few studies that are available are heterogeneous, with significant differences in study size, design, and quality for reporting of adverse events. These results should be treated with caution. Nevertheless, the evidence for harms indicate that, even though the risk is low for arthroscopic lavage, debridement and/or APM, there is still some risk associated with undergoing these arthroscopic knee procedures Cost effectiveness of arthroscopic investigation and treatment of the knee We found two relevant cost effectiveness studies. The first study [28] by Losina et al (2015) was a simulation over ten years extrapolating data from a randomised controlled trial carried out in the United States by Katz et al [29]. The Katz et al trial was included in the systematic reviews and meta-analyses by both Thorlund et al [17] and Brignardello-Petersen et al [19]. It was the only trial out of the nine RCTs included in the SRMA by Thorlund et al (2015) to show a positive effect for patient reported pain (effect size 0.22 (95%CI 0.01 to 0.44) although this was only just significant. We noted that, although chronically locked knee was an exclusion criterion for this trial, the inclusion criteria did allow for some episodic locking and catching. Since there were two other RCTs which also included some mechanical symptoms [31, 33] the positive outcome was unlikely to be due to the inclusion of episodic locking. The aim of the cost effectiveness study was to establish if arthroscopic partial meniscectomy (APM) was a cost effective treatment for symptomatic meniscal tear (excluding chronic locking) plus OA. The simulation model estimated the cost effectiveness of three different treatment

15 15 EVIDENCE SUMMARY REPORT strategies over ten years: physical therapy for three months (PT) only, a trial of PT followed by APM (delayed APM) or immediate APM. Each strategy allowed subjects to proceed to total knee arthroplasty (TKA) if they developed advance knee OA and sustained pain. The strategy was preferred if it had the highest probability of having the highest net monetary benefit of all the strategies under considerations. The net monetary benefit was the product of willingness to pay and strategy effectiveness minus the cost of delivering the treatment strategy. The model indicated that the most cost effective strategy was delayed APM ($12,000/qualityadjusted life year (QALY)(c. 9,000/QALY) compared to PT alone), and that this strategy had a 58% probability of being cost effective at a willingness to pay threshold of $50,000/QALY (c. 35,000/QALY). Immediate arthroscopic partial meniscectomy in people with symptomatic meniscal tear (excluding chronically locked knee) and knee OA was not cost effective, with an additional cost per QALY 15 of $103,000 (c. 71,000) when compared to delayed APM. Sensitivity analysis indicated that, as the effectiveness of delayed APM improved, immediate APM became less cost effective. However, as this finding was based on the only RCT in the systematic review by Thorlund et al (2015) [17], which reported a significant difference in the treatment group for pain (not function), it is highly uncertain if delayed APM is cost effective given its dependency on proven clinical effectiveness for wider outcomes as well as pain (e.g. function and quality of life). The major concern about this study was that it took into account societal costs (e.g. attached monetary value to the time subjects spent in pain, personal productivity losses) which are not included in the methodology used for deriving UK cost effectiveness estimates. This significantly lowered the reported cost per QALY as the treatment related costs were only a small proportion of the total costs subjects accumulated and the time costs more than tripled the total costs incurred over ten years. A recent project, funded by the UK NIHR and MRC Methodology Research Programme, estimated that the NICE cost effectiveness threshold should be 13,000/QALY in order to confirm that the health expected to be gained exceeds the health expected to be forgone elsewhere as other NHS activities are displaced (i.e. whether the technology is cost effective ) [30]. More recently, a cost effectiveness study was published by a Canadian group (Marsh et al 2016) [25]. The aim of this analysis was to determine the cost effectiveness of arthroscopic surgery in addition to non-operative treatments compared with non-operative treatments alone in patients with knee OA. The analysis was based on another RCT [31] which was also included in the systematic review by Brignardello-Petersen et al (2017). Some of the patients included in this RCT had mechanical locking, but the results for those with locking were not reported separately. This systematic review reported no significant difference for pain [17] between the arthroscopy and control groups. We noted that the non-operative treatment as described in the trial may not be generalisable to a UK, non-trial setting. This is because clinical trials often enjoy a higher level of compliance for conservative treatments and in real life clinical practice it is likely that the duration of physiotherapy would be shorter than 12 weeks, and access to and benefit from patient support and/or self-management groups more variable. 15 A quality-adjusted life-year (QALY) takes into account both the quantity and quality of life generated by healthcare interventions. It is the arithmetic product of life expectancy and a measure of the quality of the remaining life-years.

16 16 EVIDENCE SUMMARY REPORT Despite the inclusion of some people with mechanical locking (and therefore potential for greater benefit) and the better than usual non-operative treatments, this cost effectiveness study found that there was a less than 20% probability of arthroscopy being cost effective, even if there was a willingness to pay c. $400,000 (c. 193,000) for a clinically important improvement in WOMAC knee score or more than C$50,000 (c. 24,000) per QALY. The authors concluded that arthroscopic debridement of degenerative articular cartilage and resection of degenerative meniscal tears, in addition to non-operative treatments for knee OA is not an economically attractive treatment option compared with non-operative treatment only, regardless of willingness-to-pay value. The cost effectiveness of arthroscopic surgery is highly dependent upon the procedure s clinical effectiveness. If the intervention is not clearly clinically effective, it cannot be cost effective.

17 17 EVIDENCE SUMMARY REPORT Table 1: Summary of evidence of the clinical and cost effectiveness of arthroscopic partial meniscectomy and /or debridement Level of Evidence Study Patients Intervention Comparator Outcomes Comments Systematic Review and Meta-analysis Level 1a Brignardello- Petersen et al 2017[19] Arthroscopy vs Conservative Management Reported pain/function using hierarchy of PROMs[34] Update of SRMA by Thorlund et al 2015 SR of 13 RCTS for effects of arthroscopic surgery AND SR of 3 RCTs and 11 observational studies for complications of surgery Search date 16 Aug 2016 Symptomatic degenerative knee disease defined as persistent knee pain that affects patient s QoL and does not respond to conservative treatment with or without osteoarthritis at any age. Arthroscopic surgery including any or all of debridement and/or partial meniscectomy Conservative management strategy (exercise therapy, injections, drugs, sham surgery, no treatment) At 3 months Pain: 20 vs 15 points (MD more (95%CI 1.9 more to 8.8 more) Difference 17 : 124 more per 1000 Function: 14 vs 9 points (MD 4.9 more (95%CI 1.5 more to 8.4 more) Difference: 134 more per 1000 QoL 18 : 14 vs 8 points (MD 6.0 greater (95%CI 1.5 fewer to 13.5 more) Long term 1-2 yrs Pain at 2 yrs: 22 vs 19 points (MD 3.13 more (95%CI 0.17 fewer to 6.43 more) Function at 2 yrs: 13 vs 10 Meta-analysis required conversion of outcomes measured by different instruments to the highest ranked index instrument identified [34] Excluded studies<10 patients Complications reported at 3 months post-op Patient reported continuous measures MID 20 For complications: 16 The mean difference (MD) measures the absolute difference between the mean values of the arthroscopy versus conservative management groups. 17 Difference in number of patients who achieve a change higher than the minimally important difference (MID) 18 Quality of life (QoL) 20 Minimally important difference (MID). This was based on a systematic review and case study by Devji et al (2017) [34] and addressed what level of individual change on a given scale is important to patients (minimally important difference). The study recommended that the minimally important difference for pain and for function was 12 points and 8 points respectively.

18 18 EVIDENCE SUMMARY REPORT Level of Evidence Study Patients Intervention Comparator Outcomes Comments points (MD 3.16 more (95%CI 0.48 fewer to 6.8 more) QoL at 1 yr: 12.4 vs 10.3 points (MD 2.12 more (95%CI 0.96 fewer to 5.21 more) Knee replacement 23 vs 12 per 1000 (MD 11 more per 1000 (95%CI 107 more 6 fewer) Complications per 1000 procedures at 3 months (arthroscopy vs conservative management) risk of bias and low certainty due to: large proportion of retrospective studies patient characteristics varied in age and gender wide range of study size and design included Mortality (7 studies, n=454,086): 0 vs 0 per 1000, <1 more (95%CI 0 to 1) VTE 19 (11 studies, n=1,119,920): 5 vs 0 events per 1000, 5 more (95%CI 2 to 10) Infection (5 studies, n=603,838): 2 vs 0 events per 1000, 2 more (95%CI 1 to 4) Nerve damage (1 study, n=12,426): 0 vs 0 per 1000, <1 more (95%CI 0 to1) 19 Venous thromboembolism (VTE)

19 19 EVIDENCE SUMMARY REPORT Level of Evidence Study Patients Intervention Comparator Outcomes Comments Level 1a Reichenbach et al 2010 [18] SRMA of 7 randomised or quasirandomised trials Search date 3 August 2009 n=567 adults with knee osteoarthritis. Arthroscopic and nonarthroscopic joint lavage 3 trials examined arthroscopic joint lavage, 2 nonarthroscopic joint lavage and 2 tidal irrigation. Sham intervention, placebo or nonintervention control No significant difference for pain relief: At 3 months, SMD (95% CI to 0.21) corresponding to 3mm on a 100mm visual analogue scale (VAS) At 1 year, SMD (95%CI to 0.29) No significant difference for function: At 3 months, SMD (95% CI to 0.11), corresponding to 0.2 cm on a WOMAC disability subscale from 0 to 10 At 1 year SMD (95%CI to 0.03) The methodological quality and the quality of reporting in the RCTs was poor Moderate to large degree of heterogeneity among the trials (I 2 = 65%) For pain, estimates of effect sizes varied depending on the type of lavage, but this variation was likely to be explained by differences in the credibility of control interventions Reporting on adverse events and dropout rates was poor 21 SMD = standard mean difference

20 20 EVIDENCE SUMMARY REPORT RCTs Level 1b RCT to examine 1. Subgroups (mechanical & unstable tears) 2. Failed conservative management Sivhonen et al 2017 [21] FIDELITY trial Superiority trial for APM vs placebo over 2 years 5 centres, Finland n=146 Adults age years with knee symptoms over 3 months, degenerative medial meniscus tear and unresponsive to conservative therapy, no clinical or radiographic knee OA NB excluded trauma and recent history of locked knee APM n=70 f/up at 2,6,12,24 months by questionnaire 24month clinical assessment by independent clinician Placebo surgery n=74 At 24months Primary outcomes APM vs placebo (mean (95%CI) 22, between group difference in improvement from baseline (mean (95% CI). WOMET (75.4 to86.5) vs 86.1(80.5 to 91.8), -5.2(-13.1 to 2.7) Lysholm knee score ( 78.2 to 86.3) vs 86.5( 82.3 to 90.6), -4.3(-10.0 to 1.5) Knee Pain after exercise (1.5 to 3.1) vs 1.9(1.1 to 2.7), - 0.4(-1.5 to 0.7) Secondary outcomes APM vs placebo surgery (numbers (%)) Satisfied patients: 54 (77.1) vs 58 (78.4), p= Improved patients: 61(87.1) vs 63(85.1), p=0.812 All subjects had a BMI<30kg/m 2 Patients told that they could have reoperation if no adequate relief of symptoms at 6 months Multicentre, Randomised patients, caregiver and outcome assessor blinded Placebo-surgery 22 Values are adjusted with the baseline score, study site, age, sex and the absence or presence of minor degenerative changes on a radiograph (Kellgren-Lawrence grade 0 or 1, respectively). 23 Western Ontario Meniscal Evaluation Tool (WOMET) is a meniscus specific health related quality of life instrument validated especially for patients with degenerative meniscal tear (Sivhonen 2017). It includes 16 items addressing 3 domains: 9 items addressing physical symptoms; 4 items addressing disabilities with regard to sports, recreation, work and lifestyle and 3 items addressing emotions. WOMET scores are converted to a percentage of normal, and therefore range from with 0 indicating the most severe symptoms and 100 the absence of symptoms. 24 The Lysholm knee score was originally designed to evaluate knee function and symptoms in activities of daily living in patients with anterior cruciate ligament insufficiency. It has later been adjusted and validated for the evaluation of meniscal injuries. The Lysholm knee score is a condition-specific outcome measure with eight domains: limp, locking, pain, stair climbing, use of supports, instability, swelling and squatting. An overall score of points is calculated, 0 denoting the worst possible outcome and 100 the best outcome. 25 Knee pain after exercise (during the preceding week) was assessed on an 11 point numerical rating scale ranging from 0=no pain to 10=extreme pain