NHS-Prospero registration number 42016048592
Objective In order to make a more evidence-based selection of patients who would benefit the most from arthroscopic partial meniscectomy (APM), knowledge of prognostic factors is essential. We conducted a systematic review of predictors for the clinical outcome following APM.
Design Systematic review
Data sources Medline, Embase, Cochrane Central Register, Web of Science, SPORTDiscus, PubMed Publisher, Google Scholar
Inclusion criteria Report an association between factor(s) and clinical outcome; validated questionnaire; follow-up >1 year.
Exclusion criteria <20 subjects; anterior cruciate ligament-deficient patients; discoid menisci; meniscus repair, transplantation or implants; total or open meniscectomy.
Methods One reviewer extracted the data, two reviewers assessed the risk of bias and performed a best-evidence synthesis.
Results Finally, 32 studies met the inclusion criteria. Moderate evidence was found, that the presence of radiological knee osteoarthritis at baseline and longer duration of symptoms (>1 year) are associated with worse clinical outcome following APM. In addition, resecting >50% of meniscal tissue and leaving a non-intact meniscal rim after meniscectomy are intra-articular predictive factors for worse clinical outcome. Moderate evidence was found that sex, onset of symptoms (acute or chronic), tear type or preoperative sport level are not predictors for clinical outcome. Conflicting evidence was found for the prognostic value of age, perioperative chondral damage, body mass index and leg alignment.
Summary/conclusion Long duration of symptoms (>1 year), radiological knee osteoarthritis and resecting >50% of meniscus are associated with a worse clinical outcome following APM. These prognostic factors should be considered in clinical decision making for patients with meniscal tears.
- meniscal injury
- prognostic factors
- systematic review
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
1 For many years, arthroscopic partial meniscectomy (APM) has been considered the gold standard for torn menisci, for both traumatic and degenerative tears.1 ,2 3 Yearly, over 700 000 APMs are performed in the USA.4 Although it remains one of the most common surgical procedures in many Western countries,5 several high-quality randomised controlled trials (RCTs) challenge the indications of APM.4 6–9 These trials, summarised in a recent systematic review,10 consistently show no benefit in function and pain relief of APM compared with physical therapy or sham surgery in patients with degenerative meniscal tears. Furthermore, there is a growing concern that patients who have undergone APM are at increased risk of developing knee osteoarthritis (OA).2 11
Taking the results of the earlier mentioned RCTs and the concern about knee OA into account, a more evidence-based approach in patient selection for APM is needed. Instead of considering APM the standard of care, clinicians need to carefully select subgroup of patients with meniscal pathology who would likely benefit from APM. If one can predict the ‘chance of success’ (ie, patient-reported pain, physical function level) following APM based on patient characteristics, a more evidence-based patient selection can be made. In order to predict this chance of success, knowledge of prognostic factors is essential.
To the best of our knowledge, no systematic review of prognostic factors for the clinical outcome following APM has been conducted. We systematically reviewed all available literature, to determine the association between certain preoperative and operative variables and clinical outcome following APM. The purpose of this study was to identify prognostic factors for the clinical, patient-reported outcome of APM in patients with a meniscal tear.
The reporting in this systematic review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement.12 This study was registered in the International Prospective Register for Systematic Reviews of the National Institute for Health Research, no. 42016048592.
A health science librarian of our institution with extensive experience in the conduct of literature searching for systematic reviews assisted in designing and performing the search. We searched in Medline, Embase, Cochrane Central Register, Web of Science, SPORTDiscus, PubMed Publisher and Google Scholar for relevant articles (date of search: 16 September 2016). The following main keywords were used: knee, meniscus, meniscal tear, treatment and meniscectomy (see online supplementary appendix 1 for complete search). The articles types included in the search were RCTs and prospective or retrospective cohort studies. There was no date of publication restriction in the search.
Supplementary file 1
The inclusion criteria for the present study were: (1) all subjects had to have a meniscal tear, confirmed by MRI/arthroscopy/X-ray with contrast, treated with APM; (2) subjects had to be aged over 18 years; (3) the study had to describe a correlation/association between one or more prognostic factors and patient-reported clinical outcome (from now on described as ‘clinical outcome’) of APM; (4) a validated patient-reported outcome measure had to be used; (5) there had to be a follow-up of at least 12 months and (6) the article had to be written in English, German, Dutch, French, Spanish or Swedish. We choose these languages because members of the project group were able to read these.
We excluded studies which (1) had <20 subjects; (2) included patients with anterior cruciate ligament (ACL) deficiency or with previous ACL reconstruction; (3) included patients with discoid menisci; (4) included patients undergoing meniscal repair; (5) included meniscus transplantation or meniscus implants; (6) included patients undergoing total meniscectomy; (7) included patients undergoing open meniscectomy and (8) included additional surgical interventions carried out at arthroscopy.
Two reviewers independently screened all titles and abstracts for eligibility. Disagreements were discussed and resolved by consensus. A third reviewer was asked in case of unsolved disagreement. Duplicate studies were removed using a validated method developed by the medical library of our institution, consisting of several steps. Furthermore, reference lists of all selected studies were searched to identify potential missed articles.
Risk of bias
To assess the potential risk of bias, two reviewers independently assessed each study using the Cochrane Collaboration’s tool for assessing risk of bias of prognostic studies.13 14 This scoring list involves eight questions: two questions concerning selection bias, four questions concerning information bias and two questions concerning confounding. A low risk of bias was defined as (1) ‘yes’ to at least six out of eight questions and (2) at least one time ‘yes’ in each risk of bias category (selection bias, information bias, confounding). A moderate risk of bias was defined as (1) ‘yes’ to at least five out of eight questions and (2) at least one time ‘yes’ in two of the risk of bias categories. All other cases were considered as high risk of bias. The two reviewers discussed their findings and asked a third reviewer for consensus, if necessary.
Data regarding study design, level of evidence, number of patients, population characteristics, arthroscopic findings, outcome measurements, results and associated prognostic factors were extracted by one reviewer, using a standardised form.
Best evidence synthesis
The clinical and methodological homogeneity of the included studies was checked to evaluate whether a meta-analysis would be appropriate. If not, a best evidence synthesis was performed, using the algorithm developed by van Tulder et al.15–17 By summarising findings while taking the weight of the evidence into account in a standardised way, a best evidence synthesis provides conclusions based on the best available evidence. The following ranking of levels of evidence was used: (1) strong evidence is provided by two or more studies with low risk of bias and by generally consistent findings in all studies (≥75% of the studies reported consistent findings); (2) moderate evidence is provided by one low risk of bias study and two or more moderate/high risk of bias studies or by two or more moderate/high risk of bias studies and by generally consistent findings in all studies (≥75%); (3) limited evidence is provided by one or more moderate/high risk of bias studies or one low risk of bias study and by generally consistent findings (≥75%); (4) conflicting evidence is provided by conflicting findings (<75% of the studies reported consistent findings); (5) no evidence is provided when no studies could be found.
Besides overall analysis, subgroup analysis was performed regarding age (under and above 45 years).
We identified 5150 potentially relevant articles: 5146 by electronic search and 4 by reference tracking. After screening on title and abstract, 159 studies were considered to be potential eligible (figure 1). Full text of these studies was assessed, and 32 studies met our inclusion criteria and were included (see table 1 for study characteristics and main results).
Characteristics of included studies
We included 1 RCT,6 4 prospective follow-up studies18–21 and 27 retrospective studies. Overall, the included studies had allocated 4250 patients (range 2622–109023). The follow-up ranged from 16 20 23 24 to1325 26 years. The mean age of patients of the included studies ranged from 1925 to 6027 years. Most articles included patients with all types of meniscal tears; however, two studies28 29 only included radial tears, two studies30 31 only horizontal tears, one study32 only included root-tears, one study33 only complex tears and one study34 only bucket-handle tears. Five studies excluded patients with a certain degree of chondral damage. Furthermore, 13 studies excluded patients with knee OA (mostly based on radiographs).
Risk of bias of included studies
For 26 35 of the 32 included studies we found a low risk of bias. For the remaining studies, a moderate-to-high risk of bias was found. A risk of selection bias was found in 77% of the included studies, a risk of confounding in 94% and a risk of information bias in none of the studies. The agreement between reviewers in the risk of bias assessment was 98%.
A considerable variability was found between included studies regarding study population, the definition of subgroups and outcome measures. Furthermore, clinical outcomes of individual subgroups were often inadequately described or lacking completely. Taking the considerable heterogeneity and lacking subgroup outcomes into account, pooling data and conducting a meta-analysis was not appropriate. Hence, qualitative analyses were performed, according to the best evidence synthesis principle.
In total, 13 different prognostic factors were identified and shown to be associated with clinical outcome following APM. Table 2 shows an overview of prognostic factors, which are described in at least two studies.
Duration of symptoms
Two studies36 37 evaluated the duration of symptoms in the context of clinical outcome. In one study,36 acute (symptoms existing <12 months) and chronic (symptoms existing >12 months) lesions are distinguished, one study37 defined a duration of 3 months or less as ‘short’, and longer than 3 months as ‘long’. Both studies concluded that a shorter duration of symptoms is statistically significantly associated with better patient-reported outcome measures.
Radiological knee OA at baseline
Two studies38 39 described the presence of radiological knee OA and its association with clinical outcome of APM. In one study,39 patients with no sign of knee OA (Kellgren and Lawrence40 grade 0) and patients with mild-to-moderate knee OA (Kellgren and Lawrence grade 1–2) were included. One study38 also included patients with severe knee OA (Fairbank41 grade >2). Both studies reported a statistically significant smaller improvement of Lysholm knee scores in patients with radiological knee OA at baseline.
Amount of resected tissue
Six studies assessed the relationship between the amount of resected tissue during APM and clinical outcome. Five out of six studies reported a positive association between the amount of resected meniscal tissue and decreased patient-reported outcome measures. In two studies,25 42 a ‘subtotal’ procedure (>50% resected, leaving a small rim of meniscal tissue) was found to result in worse clinical outcome than a ‘partial’ procedure (<50% of meniscal tissue resected). Other studies described the absence of the meniscal rim43 or a preserved meniscal width of <3 mm44 as a predictor for worse clinical outcome. In one study,45 the method for measuring the influence of this factor on clinical outcome was not further described. One study,46 which investigated the influence of the percentage of removed tissue in 31 knees with lateral meniscal tears, found no association with postoperative Lysholm scores.
No prognostic factors
The influence of sex on clinical outcome after APM was assessed in 10 articles. Eight of them reported no statistically significant association between sex and outcome. Two studies19 20 reported a worse outcome for women.
The influence of onset, that is, traumatic versus non-traumatic, on outcome after APM was assessed in eight articles and seemed not to be a predictor for clinical outcome. Two studies27 47 reported a worse outcome for non-traumatic tears, based on arthroscopic findings. However, six studies reported no statistically significant correlation.
Preoperative sport level
In four studies, preoperative sport level was assessed. Two studies42 43 distinguished a recreational and competitive sport level, one study20 measured the hours of exercise per week and one study48 did not further specify study groups. None of the articles found a correlation between sport level and clinical outcome of APM.
Type of meniscal tear
In nine studies, the association between the type of meniscal tear and clinical outcome was assessed. Eight of them found no association, whereas one study36 reported a worse outcome for complex and for degenerative tears. None of the studies described a classification system used for the type of meniscal tears. Furthermore, a large variety among studies was found regarding the definition of subgroups (types of meniscal tears). The amount of subgroups ranged from two36 48 to five.23
An association between the location of the tear (medial vs lateral meniscus) and clinical outcome of APM was only described in one of our included studies43; in this study, no statistically significant difference was found between medial and lateral APMs. Regarding the side of knee,23 the location of chondral damage49 and perioperative synovial inflammation,21 no correlation with clinical outcome was found as well. Furthermore, one of the included studies20 assessed the predictive value of self-reported fitness at baseline and prior knee surgery and found a worse Lysholm score 1 year after APM for women with lower self-reported fitness. For men, no influence was found of self-reported fitness on clinical outcome. Prior knee injury resulted in a lower Lysholm after APM in women, in men however no such association was found.
Age at baseline
The influence of age on clinical outcome following APM was investigated in 11 studies. In two studies,26 48 patients were divided into two groups: aged under 30 years and aged above 30 years. One article36 divided patients in a group under and above 40 years. In the remaining studies, the method for defining age subgroups was not specified. Five studies found a worse clinical outcome for older patients, and six studies did not find a statistically significant association.
Body mass index
Seven studies described the association between body mass index (BMI) and clinical outcome. Four of them reported a worse Lysholm score for overweight or obese patients. The remaining studies found no association between BMI and clinical outcome. When we looked at studies with patients aged above 45 years, we found evidence for the fact that there is no association between BMI and clinical outcome of APM.
The predictive value of leg malalignment was described in three studies. One of them38 reported a statistical significantly worse modified Lysholm score for patients with a valgus malalignment (tibiofemoral angle more than four degrees on anteroposterior full leg radiograph). However, two studies22 43 found no significant association between leg malalignment and clinical outcome.
Chondral damage during arthroscopy
Ten studies investigated the association between chondral damage found during surgery and clinical outcome. Three of them used the Outerbridge50 classification, two of them the International Cartilage Repair Society51 classification, and the remaining studies only mentioned whether chondral damage was found during arthroscopy or not. Six out of 10 studies reported that the presence of chondral damage predicted a worse clinical outcome, and 4 studies did not find such an association. The relationship between chondral damage and clinical outcome seems to be driven by age; when we looked at studies with patients aged above 45 years (n=4), all studies reported a worse outcome for patients with chondral damage during arthroscopy. Looking at studies with patients aged below 45 years (n=6), almost all studies reported no association between chondral damage and outcome. Furthermore, when specifically looking at medial meniscal tears, chondral damage seems to be a prognostic factor for worse outcome as well.
Despite the extensive heterogeneity in study design, in the definition of subgroups and in outcome measurements, several prognostic factors were found for the clinical outcome after APM. We found moderate evidence that a larger amount of resected tissue, the presence of radiological knee OA at baseline and a longer duration of complaints were associated with a worse clinical outcome following APM. Sex, the preoperative sport level, onset (traumatic vs degenerative) and the type of meniscal tear do not seem to influence clinical outcome. It should be noted that, the phrasing ‘worse outcome’ does not necessarily mean that the outcome is unsatisfactory. It means that having a specific factor is associated with a worse patient-reported outcome compared with not having this specific factor.
To the best of our knowledge, this is the first systematic review that focuses specifically on predictors for the clinical outcome following APM. Salata et al 52 conducted a systematic review in 2010 on the radiological and clinical outcome in patients undergoing meniscectomy. The authors primarily assessed outcome measurements of APM in general, but also described some features which might influence this outcome. One of their findings was that degenerative meniscal tears are statistically significant associated with a negative postoperative outcome. This is a very relevant finding, as most APMs are performed in middle-aged and elderly patients, who typically have degenerative meniscal tears.5 53–55 The findings of Salata et al are in concordance with Englund et al 56 who found that degenerative meniscal tears result in worse clinical and radiological outcome after 16 years in 155 patients undergoing APM. By contrast, a recently published and methodologically robust study of Thorlund et al 57 reported no clinically relevant difference in patient-reported knee function and satisfaction between degenerative and traumatic meniscal tears after 12 months. This is in line with the results of the current systematic review, in which no difference in patient-reported clinical outcome between non-traumatic and traumatic tears was found as well. Thus, the predictive value of non-traumatic versus traumatic meniscal tears for the clinical outcome following APM is questionable and needs to be further unravelled.
Symptom duration is a relevant factor in APM for meniscal surgery. Although a short duration of symptoms (<6 weeks) is one of the clinical variables that orthopaedic surgeons consider to be important in surgical decision making,58 robust evidence regarding the impact of timing awaiting for APM on clinical outcome is scarce. The fact that there is no standard definition of ‘acute’ and ‘chronic’ symptoms causes a substantial amount of heterogeneity between studies, which makes them difficult to compare. Nonetheless, in the present systematic review, moderate evidence was found that a longer duration of symptoms (longer than 3–12 months) is associated with a worse clinical outcome following APM.
A third key finding of the current systematic review concerns the amount of resected meniscal tissue during arthroscopy, which appeared to be a relevant factor in predicting the clinical outcome following APM. This is not surprising, given the critical biomechanical role of the meniscus within the knee joint.59 Our study suggests that the amount of resected meniscal tissue is negatively associated with postoperative clinical outcome following APM, in concordance with Englund et al 56 and Salata et al.52 More specifically, resecting >50% of meniscal tissue, leaving <3 mm meniscal width and impairing the peripheral third (the meniscal rim) were found to be associated with worse clinical outcome. In conclusion, resecting more meniscal tissue is associated with worse clinical outcome after APM.
Whereas no association was found between meniscal tears with a non-traumatic onset (compared with traumatic tears) and a worse clinical outcome following APM, our study does show that radiological knee OA at baseline is associated with a worse clinical outcome. This is in line with the results of Kirkley et al 60 showing that arthroscopic surgery for patients suffering knee OA may not lead to satisfactory outcomes. The interesting thing is that a degenerative meniscal tear, as described earlier, does not seem to be associated with a worse clinical outcome following APM. As degenerative meniscal tears are often considered to be a signifying feature of incipient knee OA,61–63 one might expect that this type of tear, compared with other types of meniscal tears, has a negative association with clinical outcome as well. Further investigation into this topic, for example, using novel imaging techniques which provide quantitative information regarding the degree of meniscal degeneration,64 is desired.
Another relevant knee-specific factor that we studied, is chondral damage during surgery. Symptomatic degenerative meniscal tears are frequently associated with cartilage damage to the corresponding articular surfaces.65 66 In the current systematic review, conflicting evidence was found for the predictive value of chondral damage on clinical outcome after APM. However, subgroup analysis showed that, when looking at the studies in patients with a mean age of <45 years, no association was found between chondral damage and outcome. For the studies in patients with a mean age of >45 years, we did find that chondral damage at time of surgery is associated with a worse clinical outcome. A study by Sofu et al 67 in which patients aged above 60 years with traumatic meniscal tears were included, reported worse pain scores for patients with chondral damage as well. Thus, it is likely that chondral damage in patients aged above 45 years has a negative influence on clinical outcome following APM, however this association needs to be further investigated.
A factor that could potential be of influence on clinical outcome is whether the tear is located in the lateral- or the medial meniscus. However, this factor was studied in only one of the included publications, which did not find an association. As a potential prognostic factor needs to be described in at least two studies, according to the best evidence synthesis principle, no conclusions regarding the predictive value of medial versus lateral meniscectomies can be drawn. This factor is particularly relevant as in literature, lateral meniscectomy has been reported to result in poorer postoperative outcome than medial meniscectomy.52 68–70 A hypothesis is that the lateral meniscus is ‘less conforming’ than the medial meniscus after meniscectomy, resulting in an increased amount of instability and resultant force transmission to the articular cartilage. By all means, the predictive value of this factor too warrants further investigation.
A major strength of the present study is that we performed an extensive search in all relevant databases by aid of an experienced biomedical information specialist of the medical library of our institution. Furthermore, the majority of steps in this systematic review were performed in duplo, and acknowledged tools for the assessment of the risk of bias and data extraction were used. A limitation of our systematic review is that, despite the large amount of found publications, relatively few studies could be included in this systematic review. This is a consequence of our selection strategy, involving extensive exclusion criteria. To increase the a priori chance of acquiring reliable and comparable results (and potential conduct a meta-analysis), we defined concrete, well-justified and clearly stated eligibility criteria. For example, we only included articles using validated questionnaires, such as the Lysholm or International Knee Documentation Committee71 score. Publications using outcome measures such as ‘percentage of satisfied patients’ were therefore excluded. The rationale of this exclusion criterion is the relatively low reliability and reproducibility of non-validated patient-reported outcome measurements. Although we might have missed information about prognostic factors, we believe that this approach increased the reliability of our results.
Another limitation of this systematic review is, that only rough estimations of the effect size of the found prognostic factors could be provided. This is due to the fact that a substantial amount of heterogeneity in the definition of subgroups and outcome measurements was found. For example, the potential influence of the type of meniscal tear on clinical outcome following APM was reported in nine studies; however, none of them described a classification system for the type of tear. In fact, six of them did not provide any information regarding the definition of meniscal tear subgroups at all. Also, in many of the included studies the outcome of subgroups was poorly described. Often only P values were reported; some studies did not even provide a P value but only described the prognostic value of a specific factor (eg, ‘No significant correlation was found between the amount of tissue resected and the subjective, clinical and radiological outcome’).46 Given the found heterogeneity and inadequately described subgroup results, pooling of study results and performing a meta-analysis were not justified. This implied that small studies might not have reported an association based on lower power while pooled results the reported association would have counted in the overall estimation for the association. By summarising findings while taking risk of bias into account, a best evidence synthesis provided conclusions based on the best available evidence. Given that most studies in the present systematic review showed a high risk of bias, only moderate and limited evidence for prognostic factors could be provided.
Despite the high amount of APMs performed worldwide, there is a lack of consensus on the indications for this procedure, particularly in younger and middle-aged patients. To enable a more evidence-based approach in surgical decision making, knowledge of the predictive value of certain patient-specific factors for the clinical outcome is essential. In this comprehensive systematic review, prognostic factors for the patient-reported outcome of APM were assessed. We have shown that based on the best available evidence, radiographic knee OA at baseline, a long duration of complaints and resecting more meniscal tissue during arthroscopy are associated with a worse postoperative clinical outcome. The findings could contribute to the development of a prediction model for the clinical outcome of APM, based on patient-specific factors, which could guide orthopaedic surgeons in their clinical decision making. However, within the available literature, the earlier mentioned heterogeneity and inadequately reported subgroup outcomes make it challenging to draw adequate conclusions. Therefore, there is an urgent need for more well-designed, robust clinical trials on arthroscopic meniscal surgery using validated patient-reported outcome measurements and with relevant, a priori defined subgroups. These subgroups may include a standardised and solid classification of meniscal tear type, and a standardised way of defining and classifying the duration of symptoms.
What are the new findings?
This is the first review that focuses specifically on predictors for the clinical outcome of arthroscopic partial meniscectomy (APM).
The presence of radiological knee osteoarthritis (OA) and a long duration of symptoms are patient-related predictors for a worse clinical outcome after APM.
Resecting >50% of meniscal tissue, a non-intact meniscal rim after meniscectomy, and preserving <3 mm meniscal width are intra-articular prognostic factors for a worse clinical outcome after APM.
How might it impact on clinical practice in the near future?
In order to make an evidence-based selection of patients who would benefit the most from APM, knowledge about the predictive value of specific patient characteristics is essential.
The patient-related prognostic factors, found in this study (longer duration of symptoms, presence of radiological knee OA), should be considered in clinical decision making for patients with meniscal tears.
The intra-articular prognostic factors, found in this study (resecting more meniscal tissue) should be considered during arthroscopic surgery in patients with meniscal tears.
Additional methodologically robust studies are needed on arthroscopic meniscal surgery using validated outcome measurements and with adequate subgroups.
The authors would like to thank Wichor Bramer (Medical Library, Erasmus MC University Medical Centre Rotterdam, The Netherlands) for assistance with designing and performing the literature search.
Contributors SME, MR, SMAB-Z and DEM conceived and designed the study. SME, DTvY and DEM performed the literature search and selected the studies. SME collected the data from all included studies. Risk of bias was assessed by SME and MR. Study outcomes were summarised and analysed by SME, MR, SMAB-Z and DEM. SME wrote the initial draft of the manuscript, and MR, DTvY, SMAB-Z and DEM made significant revisions and contributions. All authors read and approved the final manuscript.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.