Article Text
PDF
Abstract
Objective Determine the long-term health-related quality-of-life (HRQoL), work limitation, physical activity, health/economic cost and disease burden of traumatic ACL and/or meniscal injury. Findings will inform OPTIKNEE evidence-based consensus recommendations.
Design Random-effects meta-analysis evaluated HRQoL (SF-36/SF-12/VR-12 Physical Component Scores (PCS) and Mental Component Scores (MCS), EuroQol-5D (EQ-5D)) stratified by time postinjury, and pooled mean differences (95% CI) between ACL-injured and uninjured controls. Other outcomes were synthesised descriptively. Risk-of-bias (RoB) and certainty of evidence (Grading of Recommendations Assessment, Development and Evaluation) were assessed.
Data sources MEDLINE, EMBASE, CENTRAL, SPORTDiscus, CINAHL searched inception: 22 November 2021.
Eligibility Studies reporting HRQoL, work limitations, physical activity levels, health/economic costs or disease burden, ≥2 years post-ACL and/or meniscal injury.
Results Fifty studies were included (10 high-RoB, 28 susceptible-to-some-bias and 12 low-RoB). Meta-analysis (27 studies, very low certainty of evidence) estimated a pooled mean (95% CI) PCS of 52.4 (51.4 to 53.4) and MCS of 54.0 (53.0 to 55.0) 2–14 years post-ACL injury. Pooled PCS scores were worse >10 years (50.8 (48.7 to 52.9)) compared with 2–5 years (53.9 (53.1 to 54.7)) postinjury. Excluding high-RoB studies, PCS scores were worse in ACL-injured compared with uninjured controls (−1.5 (−2.9 to –0.1)). Six studies (low certainty of evidence) informed a pooled EQ-5D score of 0.83 (0.81 to 0.84). Some individuals experienced prolonged work absenteeism and modified activities ≥2 years post-ACL injury. ACL injury was associated with significant direct and indirect costs, and early ACL reconstruction may be less cost-effective than rehabilitation. Only three studies evaluated meniscal injury outcomes (all evaluated HRQoL).
Conclusion There is a very-low certainty of evidence that PCS scores ≥2 years post-ACL injury are worse than uninjured controls and decline over time, whereas MCS scores remain high. ACL injury can result in prolonged work absenteeism and high health/economic costs. Further studies are needed to determine the long-term burden of traumatic meniscal injury.
- physical activity
- osteoarthritis
- rehabilitation
- anterior cruciate ligament
- quality of life
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN?
ACL and meniscal injury are associated with an elevated risk of knee osteoarthritis as early as 10 years after injury.
ACL injury can result in knee pain, reduced knee function, fear of reinjury, sport cessation and poor knee-related quality-of-life.
The long-term burden of ACL injury on generic health constructs, including physical activity, work limitations, health/economic costs, burden of disease and overall health-related quality-of-life (HRQoL) is less clear.
The literature on traumatic meniscal injury is sparse, systematic reviews are needed to synthesise existing evidence and identify knowledge gaps.
It is unclear how generic health constructs compare to uninjured controls or population norms, ≥2 years following traumatic ACL and/or meniscal injury.
WHAT ARE THE NEW FINDINGS?
Physical aspects of HRQoL ≥2 years after ACL injury were worse than uninjured controls and declined over time, whereas mental aspects of HRQoL remained high ≥2 years after ACL injury.
Some individuals experience a prolonged period of leave from work after ACL injury, and others reduce work intensity or report work limitations ≥2 years after ACL injury.
Although people often change the type of activities they participate in after ACL injury, on average, self-reported physical activity levels may be similar to the general population. Research using objective measures of physical activity at a variety of timepoints after injury is needed.
Two randomised controlled trials found that early ACL reconstruction may be less cost-effective compared with rehabilitation and optional delayed ACL reconstruction.
There is a need for high-quality studies investigating the long-term burden of traumatic meniscal injury.
Background
After ACL injury, most individuals complete rehabilitation and are cleared to return to sport and other activities within 1 year of injury.1 Despite this, persistent knee problems are common, including knee pain, fear of reinjury, functional restrictions and poor knee-related quality of life (QoL).2–5 Only 55% of people return to competitive sports following ACL injury6 and people who do not return may become inactive and experience negative impacts on mental health and knee-related QoL.7 8 Furthermore, following ACL injury individuals have four times higher odds of developing knee osteoarthritis, and this increases to six times higher odds when combined with meniscal injury.9 Clearly, ACL injury can have long-term health impacts. However, most ACL injury systematic reviews focus on knee-specific outcomes rather than generic health measures that can be compared with uninjured populations.10 The impact of ACL injury on health-related QoL (HRQoL), work limitation, physical activity, health/economic costs and burden of disease (eg, disability-adjusted life years (DALY), years lived with disability (YLD), mortality) is poorly understood.
Traumatic meniscal injury can also impact long-term health. The meniscus plays a vital role in maintaining knee function,11 and there is a strong relationship between traumatic meniscal injury and knee osteoarthritis.9 12 However, there is a paucity of studies and very few systematic reviews evaluating long-term outcomes following traumatic meniscal injury. Understanding the long-term burden of ACL and meniscal injury could inform new strategies to improve knee injury outcomes, and guide research priorities. The primary aim of this systematic review was to synthesise the evidence on HRQoL, work limitations, physical activity, health/economic costs and burden of disease ≥2 years following ACL and/or traumatic meniscal injury. The secondary aim is to determine the burden of living with knee symptoms and osteoarthritis after ACL and/or traumatic meniscal injury by synthesising these measures in people with knee symptoms and/or structural osteoarthritis. This systematic review is one of several contributing to the development of evidence-based consensus recommendations for rehabilitation to optimise health and prevent post-traumatic osteoarthritis following knee trauma (OPTIKNEE). We focused on ACL and/or meniscal injury in this review, as these injuries are associated with the greatest risk of post-traumatic knee osteoarthritis.12
Methods
This review was conducted according to the guidelines presented in the Cochrane Handbook13 and reported following the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines.14 The protocol was prospectively registered on Open Science Framework (https://osf.io/zwty5/).
Patient and public involvement
Two individuals with lived experience of ACL tear and ACL reconstruction (ACLR) and four clinicians (ie, physiotherapists and orthopaedic surgeons) contributed to the priority theme setting of this review at an in-person meeting to plan the OPTIKNEE consensus. These individuals represent the patient and clinician perspective and were not clinician scientists or academics (in this field of study).
Search strategy
On 22 November 2021, we searched Medline (OVID), EMBASE (OVID), CENTRAL (OVID), SPORTDiscus (EBSCO) and CINAHL (EBSCO). The search strategy comprised four concepts: (i) population; (ii) outcome variable(s); (iii) follow-up and (iv) limits and exclusions. The full search strategy for each database is presented in online supplemental appendix 1. All searches were documented and records were uploaded to Covidence and autodeduplicated. The search strategy was finalised with a librarian scientist (AH).
Supplemental material
Records were randomly and equally distributed among four pairs of reviewers (n=8 reviewers), following a team meeting to ensure familiarity of eligibility criteria and consistency of application. Titles and abstracts were reviewed to identify potentially eligible studies, by at least two coauthors independently. After a consensus meeting and use of a third reviewer (SRF/CE) to resolve discrepancies, full-texts were reviewed and assessed for final eligibility with the same third reviewer used to reach consensus if necessary. Reference lists of included studies were scanned for additional articles that may be eligible.
Selection criteria
To be eligible for inclusion, articles needed to be published in English and include original quantitative data (excluding conference abstracts) from human subjects reporting eligible burden construct(s) ≥2 years post-ACL rupture (isolated or with concomitant injury (ie, meniscal injury, cartilage lesion and other ligamentous injury)) or isolated traumatic meniscal injury. Eligible constructs for burden were as follows: physical activity level, work limitations, work productivity, income, early retirement, economic cost/evaluation, healthcare costs, HRQoL, mortality, years-of-life-lost, YLD, DALY, quality-adjusted-life-year (QALY). We excluded studies not reporting (or data for calculation of) time from injury to outcome measurement (ie, studies that only reported time since knee surgery without reporting time from injury to surgery were ineligible) because the exposure of interest was ACL and/or meniscal injury and this information was required to assess time since injury for data analysis.
Data collection
Data were independently extracted and collated by the same four pairs of reviewers (n=8 reviewers), using a structured data extraction form (see protocol https://osf.io/zwty5/). Data collection forms were compared during a consensus meeting between the two reviewers. All forms were then cross-checked by an independent reviewer and any discrepancies were resolved through consensus.
Studies that only reported outcome data in figures were excluded if the data could not be confidently extracted from the figure. Where original data were presented only in figures, two independent reviewers attempted to extract the data using online software (WebPlotDigitizer V.4.2, San Francisco, CA, USA). If <1 unit variation existed in all data points between reviewers, the data were included in our review. If mean scores were reported for an eligible outcome without a measure of variability (ie, no SD or 95% CI),15–19 the SD was estimated using the median SD from all other included studies. If only 95% CIs were reported, SDs were estimated using the square root of the sample size and corresponding t-scores.20 If time since injury was reported as time from injury to surgery, and time from surgery to follow-up, mean/median values were combined to estimate time since injury for use in meta-analysis. When possible, values were converted to the same unit of measurement for analyses. If a study outcome was reported for separate eligible subgroups (eg, male and female), the subgroups were combined using a formula from The Cochrane Handbook for Systematic Reviews of Interventions to obtain mean (SD) for the combined cohort.20
Outcome measures
Common outcome measures that were reported by ≥5 studies, included versions of The Optum SF Health Surveys (SF-3621 and SF-12),22 enabling calculation of two summary scores: Physical Component Score (PCS) and Mental Component Score (MCS). Norm-based scoring is recommended for these measures, which employs a linear T-score transformation with a mean (SD) of 50 (10), derived from 1998 US general population norms.23 24 For PCS and MCS scores, group mean scores <47 can be interpreted as being below the average range for the general population.24 Another common HRQoL measure was the EuroQol-5D (EQ-5D), which evaluates five items (mobility, self-care, usual activities, pain/discomfort and anxiety/depression) which are summed to provide an overall health-status score. Weighted scores are used to calculate a utility index, where 0 represents ‘death’ and 1 represents ‘perfect health’.25
Risk-of-bias assessment
The same four pairs of reviewers independently assessed the risk of bias (RoB) for each study using the National Institute of Health Quality Assessment Tools (NIHQAT).26 The NIHQAT assists reviewers in critically appraising internal validity and includes tools specific to different study designs.26 The tools evaluate potential sources of bias (eg, selection, performance, detection, attrition, power and confounding). Reviewers consider the potential RoB associated with a specific flaw in design or implementation, and assign an overall ‘good’, ‘fair’ or ‘poor’ quality rating to each study. Detailed instructions accompanying each tool outline the importance of subjectively assessing each study on its own based on the details that are reported and consideration of bias, rather than using specific criteria to assign a rating for all studies.26 A poor rating indicates ‘high RoB’, a fair rating is ‘susceptible to some bias’ deemed not sufficient to invalidate its results, and a good rating has a ‘low RoB’ and results are considered valid.26 Further detail on the NIHQAT is available online: https://www.nhlbi.nih.gov/health-topics/study-quality-assessment-tools.
Certainty of evidence
Two authors (SRF and CBJ) assessed the certainty of evidence for each outcome according to the approach from the Grading of Recommendations Assessment, Development and Evaluation (GRADE) working group.27 Agreement was reached by consensus. The starting certainty of evidence was rated as ‘low’ when a majority of studies for a given outcome were observational. Downgrading was based on high RoB, inconsistency, indirectness, imprecision of estimates and publication bias.27 Low certainty of evidence suggests that further research is very likely to have an important impact on the confidence in the estimate and is likely to change the estimate.27 Very low certainty of evidence suggests high uncertainty about the estimate.27 An overview of the grading is provided in online supplemental appendix 2.
Supplemental material
Data synthesis
Data were synthesised descriptively if <5 studies reported the same outcome. A random-effects meta-analysis was performed when ≥5 studies reported the same outcome (meta-analysis was possible for MCS, PCS and EQ-5D scores). Pooled mean (95% CI) scores were estimated for all studies and within subgroups categorised by time since injury (ie, 2–5, >5–10 and >10 years). The Q-test based on analysis of variance (alpha<0.05) was used to compare pooled scores between time-since-injury subgroups.28 We allowed data from the same participants to be pooled in different follow-up time-points, however if data from the same participants fell within the 2–5, >5–10 or >10 years category, only the data with the greatest sample size was included in meta-analysis.
Mean differences and 95% CIs were pooled using random-effects meta-analysis in all studies that reported MCS/PCS scores in an ACL-injured and uninjured control group or a population-based sample. Heterogeneity was assessed through the Jackson method, and I2 value (high heterogeneity was considered likely when I2>50% and/or p<0.01). Publication bias was assessed through funnel plots and the Egger test. Sensitivity analyses excluding high RoB studies from each random-effects model was performed to evaluate the robustness of the results. All meta-analyses were performed in R using the meta package.29 When articles reported individual SF-36 domain scores but did not report the MCS or PCS, an R29 algorithm (R/SF-36.R) was used to convert domain scores to standardised MCS and PCS scores, using US population normative data (n=2393). If an article reported non-standardised MCS and PCS scores and did not report domain scores, these data were excluded from meta-analysis.30 31
Results
Study and participant characteristics
From the initial search identifying 2932 records, we included 50 studies (47 included ACL injured participants, 2 included participants with isolated traumatic meniscal injury, 1 included participants with ACL injury and participants with isolated meniscal injury; figure 1). Study and participant characteristics are reported in online supplemental appendix 3; 42 studies included a measure of HRQoL, 10 studies evaluated a work-related construct, 2 assessed physical activity levels and 3 studies evaluated health/economic costs and/or burden of disease. Five studies evaluated the relationship between an eligible construct and the presence of knee symptoms and/or structural osteoarthritis after ACL injury. Three studies reported an eligible construct after meniscal injury, and all three evaluated HRQoL. A total of 11 875 participants were included and follow-up ranged from 2 to 38 years postinjury.
Supplemental material
PRISMA 2020 flow diagram outlining search and eligibility screening outcomes. PRISMA, Preferred Reporting Items for Systematic reviews and Meta-Analyses
Risk-of-bias assessment
Of 23 cohort or cross-sectional studies, 2 were low-RoB,32 33 17 were susceptible-to-some bias8 15 34–48 and 4 were high-RoB19 49–51 (online supplemental appendix 4). Common sources of bias included insufficient study power, loss to follow-up and not accounting for confounding. Of 15 randomised controlled trials (RCTs), 9 were low-RoB30 31 52–58 and 6 were susceptible-to-some bias16 17 59–62 (online supplemental appendix 4). Common sources of bias included a lack of blinding, no allocation concealment, insufficient sample size or inappropriate reporting of sample size. Of eight pre-experimental or before–after studies with no control group, four were susceptible-to-some bias,63–66 four were high-RoB67–70 (online supplemental appendix 4) and most had an insufficient sample size. Three case series studies were of low-RoB,71 susceptible-to-some bias72 or high-RoB18 and one case–control study was high-RoB73 (online supplemental appendix 4).
Supplemental material
ACL injury
Health-related quality of life
HRQoL was evaluated with the SF-36,16 17 19 30 32 36 37 44 49 50 52 53 57 59–61 64–66 70 72–74 SF-12,15 18 33 38 43 56 68 Veterans RAND 12-Item Health Survey (VR-12),45 67 71 EQ-5D34 39 42 55 64 69 and AQoL-8D.8 35
Physical component scores of the SF-36, SF-12 and VR-12
Meta-analysis (27 studies) resulted in a pooled PCS score (95% CI) of 52.4 (51.4 to 53.4) (figure 2). Pooled PCS scores were similar after excluding high RoB studies18 19 49 50 67 70 73 52.0 (50.8 to 53.3), 21 studies). Stratification of PCS scores based on time since ACL injury demonstrated group differences (Q=11.0, p=0.004) (pooled PCS mean (95% CI) 53.9 (53.1 to 54.7) at 2–5 years (13 studies), 51.4 (49.4 to 53.4) at >5–10 years (nine studies) and 50.8 (48.7 to 52.9) at >10 years (six studies)) (figure 2). Pooled data from seven studies reporting PCS scores in ACL-injured and non-injured groups or population norms revealed no between group difference (mean difference (95% CI) −0.2 (−2.0 to 1.6)) (figure 3). This difference increased after removal of two high RoB studies,50 70 where PCS scores were better in healthy controls compared with ACL-injured groups (−1.5 (−2.9 to –0.1), 5 studies).
Random-effects meta-analysis of SF-36, SF-12 and VR-12 Physical Component Scores, stratified by time since ACL injury. Mean (boxes) and 95% CIs (whiskers) are presented for individual studies and the pooled mean and 95% CI are represented by the diamonds; MRAW=raw means, that is, untransformed. The norm-based scoring algorithm employs a linear T score transformation with a mean of 50 and a SD of 10, group mean scores below 47 can be interpreted as being below the average range for the general US population; green shading=low risk of bias; orange shading=susceptible to some bias; red shading=high risk of bias. PCS, Physical Component Scores.
Forest plot of random-effects meta-analysis of PCS scores in ACL injured and healthy controls (non-injured or population norms). Mean differences (boxes) and 95% CIs (whiskers) are presented for individual studies and the pooled mean difference and 95% CI are represented by the diamond. A positive mean difference indicates a favourable PCS score for ACL-injured participants; green shading=low risk of bias; orange shading=susceptible to some bias; red shading=high risk of bias. PCS, Physical Component Score from the SF-36 of SF-12.
Mental Component Scores of the SF-36, SF-12 and VR-12
Meta-analysis (27 studies) resulted in a pooled MCS score (95% CI) of 54.0 (53.0 to 55.0) (figure 4) and this was similar after exclusion of high RoB studies18 19 49 50 67 70 73 (54.3 (53.1 to 55.5)). Stratification based on time post-ACL injury resulted in similar scores (Q=4.5, p=0.11) between groups (pooled MCS score (95% CI) of 55.0 (53.8 to 56.2) at 2–5 years (13 studies), 52.5 (50.6 to 54.5) at >5–10 years (nine studies) and 54.2 (52.5 to 55.9) at >10 years (six studies)) (figure 4). Pooled data from seven studies reporting MCS scores in ACL-injured and non-injured groups or population norms revealed no between group difference (mean difference (95% CI) 0.6 (−0.9 to 2.2)) (figure 5). This finding was similar after removal of two high RoB studies50 70 (1.0 (−0.8 to 2.9)).
Random-effects meta-analysis of SF-36 and SF-12 Mental Component Scores, stratified by time since ACL injury. Mean (boxes) and 95% CIs (whiskers) are presented for individual studies and the pooled mean and 95% CI are represented by the diamonds; MRAW=raw means, that is, untransformed. The norm-based scoring algorithm employs a linear T score transformation with a mean of 50 and a SD of 10, group mean scores below 47 can be interpreted as being below the average range for the general US population; green shading=low risk of bias; orange shading=susceptible to some bias; red shading=high risk of bias. MCS, Mental Component Score.
Forest plot of random-effects meta-analysis of MCS scores in ACL injured and healthy controls (non-injured or population norms). Mean differences (boxes) and 95% CIs (whiskers) are presented for individual studies and the pooled mean difference and 95% CI are represented by the diamond. A positive mean difference indicates a favourable MCS score for ACL-injured participants; green shading=low risk of bias; orange shading=susceptible to some bias; red shading=high risk of bias. MCS, Mental Component Score from the SF-36 of SF-12.
EQ-5D utility scores
EQ-5D utility scores from six studies were available for meta-analysis, resulting in a pooled EQ-5D score (95% CI) of 0.83 (0.81 to 0.84) (figure 6), and this was similar after excluding one high RoB study69 (0.82 (0.81 to 0.84)). Participants had a pooled EQ-5D score of 0.82 (0.80 to 0.84) at 2–5 years, 0.85 (0.82 to 0.88) at >5–10 years and 0.86 (0.82 to 0.90) at >10 years post-ACL injury (figure 6).
Pooled EQ-5D utility scores stratified by time since ACL injury. Mean (boxes) and 95% CIs (whiskers) are presented for individual studies and the pooled mean and 95% CI are represented by the diamonds; green shading=low risk of bias; orange shading=susceptible to some bias; red shading=high risk of bias. EQ-5D, EuroQol-5D; MRAW, raw means, that is, untransformed.
Work-related constructs
Ten studies evaluated work-related constructs following ACL injury.8 41 46 48 54 55 58 62 66 75 Six studies assessed total days of work absenteeism at least 2 years following ACL surgery. Meta-analysis was not performed due to sample heterogeneity; two studies evaluated work absenteeism for people on work compensation claims41 75 and these studies reported longer work absenteeism than other studies (mean 11975 and 34441 days). Other studies reported the following days of work absenteeism: median (IQR) 7 (3–14) (early ACL repair) and 9 (7–19) (delayed ACL repair),48 mean (SD) 65 (88) range 5–442 (double-bundle ACLR) and 80 (83) range 5–284 (single-bundle ACLR),55 mean (SD) 77 (28),46 mean (95% CI) 86 (70 to 102).66 At a mean 27 months after ACL injury, 2 (11%) individuals on work-compensation were unable to work due to their knee, 6 (32%) were working with knee limitations and symptoms and 2 (11%) had reduced their occupational work intensity (compared with 0 (0%), 2 (11%) and 1 (5%) participant(s) in the non-work cover group, respectively).41 A study where 78% of ACL injured participants were employed in the military found that 13% had not returned to work 2 years after ACLR.62 Filbay et al (2017) reported a mean (SD) Workplace Activities Limitations Scale score of 4 (3) (scores range from 0 to 36 (higher score=greater impairment)) in 162 individuals with knee symptoms 5–20 years post-ACLR.8
A Swedish RCT reported a production loss cost of €8145 SD 9373 (double-bundle ACLR) and 9063 SD 7993 (single-bundle ACLR), per patient.55 An analysis of 5-year follow-up data from the KANON RCT found that early ACLR resulted in a mean productivity loss of €12 355 compared with €9476 following rehabilitation and optional delayed ACLR.54 Analysis of 2-year data from the COMPARE RCT found that early ACLR resulted in a mean productivity costs of €8489 compared with €7214 following rehabilitation and optional delayed ACLR.58
Physical activity
Two studies evaluated self-reported physical activity levels ≥2 years post-ACL injury.51 63 Seven-day physical activity was similar >20 years post-ACL injury after ACLR (median (range) 1563 (480–7572) MET-min/week), management with physiotherapy-alone (1217 (212–7398) MET-min/week) and healthy controls (1893 (499–8958) MET-min/week).51 Between 2 and 3 years post-ACL injury, each week participants spent a mean (95% CI) 453 (379 to 527) minutes walking, 321 (274 to 368) minutes performing household activities, 214 (183 to 246) minutes performing physical activities classified as a lower knee injury risk (eg, golf, cycling and swimming) and 92 (58 to 125) minutes in higher knee injury risk activities (eg, football, snowboarding and basketball).63
Health/economic costs and burden of disease
Health/economic costs and burden of disease were evaluated in three studies.54 55 58 An RCT estimated the total costs (direct and indirect) 26 months post-ACLR at a mean€13 718/patient for double-bundle and €14 913/patient for single-bundle ACLR.55 QALYs (reflecting EQ-5D change from baseline to 2 years) for the single-bundle and double-bundle group were 1.44 and 1.43, respectively.55 There was a 50% probability of double-bundle ACLR being cost-effective at a threshold of €50 000.55 Using KANON RCT data collected over 5 years, the mean cost of early ACLR was €4695 higher than rehabilitation and optional delayed ACLR and provided an additional 0.13 QALYs.54 The mean discounted 5-year cost for a patient in the early ACLR group was €26 200 compared with €21 478 in the optional delayed ACL reconstruction group.54 A RCT in the Netherlands found that it takes €48 460 from a healthcare perspective and €78 179 from a societal perspective to gain a QALY when performing early ACLR compared with rehabilitation and optional delayed ACLR.58 Healthcare system costs were €6368 (SD 1630) following early ACLR and €4267 (SD 3011) following rehabilitation and optional delayed ACLR. Early ACLR was not determined to be cost-effective compared with rehabilitation and optional delayed ACLR.58
The impact of knee symptoms and/or structural osteoarthritis on knee injury burden
People with greater levels of knee pain reported worse SF-36 Bodily Pain, General Health, Mental Health, Physical Functioning and Social Functioning domain scores 7–8 years post-ACL injury compared with ACL-injured people with less/no knee pain.73 Von Porat et al (2004) found similar SF-36 domain scores in male soccer players 14 years post-ACL rupture with (n=50) and without (n=72) radiographic osteoarthritis.44 Patients with a history of ACL injury scheduled for total knee replacement (TKR) reported mean (SD) PCS of 34.4 (6.2) and MCS of 39.9 (10.0), compared with PCS of 36.7 (7.0) and MCS of 43.3 (11.1) reported by patients scheduled for TKR without ACL injury history.38 Filbay et al (2018) found worse HRQoL in people with knee symptoms plus radiographic osteoarthritis (AQoL-8D median (IQR) 0.80 (0.66–0.91) or knee symptoms without radiographic osteoarthritis (0.79 (0.73–0.91)) compared with people without knee symptoms >5 years post-ACL injury (0.93 (0.88–0.97)).35 People with no/low radiographic osteoarthritis (Kellgren and Lawrence grade 0–1) post-ACL injury reported similar physical activity levels (median (IQR) 1356 (594–5295) MET-min/week), to people with moderate/high radiographic osteoarthritis (Kellgren and Lawrence grade 2–4) (1971 (212–7572) MET-min/week).51
Meniscal injury
Three studies included participants ≥2 years postmeniscal injury and all collected SF-12 data. There was no between-group differences in SF-12 scores 2–6 years postsurgery in people with bucket-handle tears who underwent isolated meniscal repair (mean (SD) PCS 53.2 (8.4), MCS 54.8 (6.0)) versus meniscal plus concomitant ligament surgery ((PCS 54.4 (5.8), MCS 55.6 (5.6)) or in people with vertical meniscal tears who underwent isolated meniscal repair ((PCS 52.6 (9.5), MCS 50.1 (9.5)), versus meniscal plus concomitant ligament surgery ((PCS 53.1 (9.1), MCS 53.8 (9.9)).40 Rodriguez-Roiz et al (2020) compared SF-12 scores 2–13 years after surgical treatment for isolated meniscal injury versus meniscal injury and concomitant ACL injury.43 Similar PCS (mean (SD) isolated meniscal: 52.5 (8.5), concomitant ACL: 52.6 (7.0)) and MCS (isolated meniscal: 55.8 (8.1), concomitant ACL: 54.1 (7.8)) were reported between groups. Scanzello et al (2013) reported SF-12 scores 2 years after meniscal injury (mean (SD) PCS 57.0 (2.8), MCS 54.2 (6.9)).68 No studies evaluated work-related constructs, physical activity, health/economic costs or disease burden measures ≥2 years following traumatic meniscal injury.
Certainty of evidence
GRADE summary of findings tables are presented in online supplemental appendix 2. Certainty of evidence was ‘very low’ for pooled PCS and MCS estimates (due to inconsistency) and ‘low’ for EQ-5D scores (no downgrades made). A ‘very low’ rating was assigned to pooled mean difference estimates for the PCS (due to risk of bias, inconsistency and imprecision) and MCS (inconsistency and imprecision).
Discussion
Health-related quality-of-life
After excluding high RoB studies, physical components of HRQoL were impaired following ACL injury compared with uninjured controls or population norms. A PCS mean difference of 1.5 points is unlikely to be clinically meaningful, although interpretation data is lacking for use in individuals with ACL-injury ≥2 years previously. Mental components of HRQoL 2–14 years following ACL injury were similar to uninjured controls or population norms. Only two studies evaluated HRQoL between ACL injured (soccer players) and uninjured participants beyond 10 years, both of which reported worse PCS scores and better MCS scores 12–14 years post-ACL injury compared with population norms.32 44 Symptomatic osteoarthritis may become more prevalent beyond 10 years of injury76 which might explain the decline in PCS scores overtime. The high MCS scores may be explained by the positive mental health impacts of sport participation (typical mechanism of ACL injury), including high resilience and pain coping strategies.77 Supporting this notion, a systematic review summarising SF-36 scores in former athletes found similar PCS scores and better MCS scores compared with population norms.78 Another consideration is that physical health may explain up to 65% of variance in MCS scores79 and the orthogonal-factor analytic model that is most commonly used to calculate summary scores can contribute to inflation of the MCS in individuals with reduced physical function.80
Knee-related QoL, which assesses knee-related impacts on QoL rather than wider health-related impacts, is impaired in the long-term after ACL injury.2 3 81 It is possible that PCS scores are more sensitive to knee-related issues (eg, knee pain, knee functional impairments) than MCS scores. Additionally, the MCS may not evaluate knee-specific mental constructs important and relevant to ACL-injured individuals, including fear of reinjury, fear of contact sports and unfulfilled competitive needs.35 Furthermore, no studies compared EQ-5D scores between ACL-injured and non-injured groups, or population norms. Based on population norms from Sweden and Norway,82 83 it is likely that EQ-5D scores are lower in ACL injured groups but higher than those reported by people with osteoarthritis (mean (SD) 0.46 (0.32),84 0.53 (0.29)85). This suggests that EQ-5D scores may reduce overtime when osteoarthritis becomes more prevalent. However, we were unable to assess this as only one study examined EQ-5D scores >10 years postinjury.
Work-related constructs
Evidence from six studies suggests that the duration of work absenteeism may vary greatly among ACL-injured samples, ranging from as low as a median 7 days to as high as an average of 344 days. Additionally, a Swedish RCT reported between 5 and 442 days of sick leave in young, active individuals following ACLR.55 The two studies with the longest work absenteeism assessed this in people on work compensation claims which may not represent the wider ACL-injured population. One study reported few work limitations on average, 5–20 years after ACLR in people with knee symptoms.8 However, 5–20 years after ACLR people could have changed occupations or modified duties to those with lower knee demands, although this was not assessed in this study. Work-related constructs may be influenced by the environment and occupation type (eg, US labourers are typically ineligible for paid time-off due to injury),86 as well as the age of study participants. More studies are needed to understand how the impact of ACL injury on work may differ between countries, occupations and age groups. The concern is those who are out of work for long periods of time or are not able to return to work.41 55 To reduce the burden of knee injury, it is important to identify risk factors for extended work absenteeism or non-return to work after ACL injury.
Physical activity
While many studies measure knee-specific activity limitations following ACL injury (eg, Tegner Activity Scale), only two studies assessed physical activity levels ≥2 years post-ACL/meniscal injury and both used self-reported questionnaires.51 63 Tengman et al (2014) found similar self-reported physical activity at 23 years post-ACL injury compared with age-matched and sex-matched controls.51 However, previously injured individuals did have lower activity on the Tegner Activity Scale, suggesting they may have adapted their participation to focus on lower impact activities. Gignac et al (2015) also reported a reduction in activities deemed ‘high risk of knee injury’ up to 3 years post-ACL injury.63 Participation in ‘low risk of injury’ activities was also reduced at follow-up but not compared with a control group. These findings align with a systematic review demonstrating that only 55% of people return to competitive sports after ACLR.6 Although Bell et al (2017) was not eligible for inclusion in our review, this study found reduced physical activity levels and step count (assessed with accelerometry) 6–67 months after ACLR compared with healthy controls.87 Given the often poor correlation between self-reported and objectively measured physical activity (eg, accelerometry), objective assessment of long-term physical activity to determine if an injury impacts an individual’s ability to meet physical activity guidelines88 is a priority.
Health economic costs and burden of disease measures
The three studies that assessed the costs of ACL injury were all based in Europe, and demonstrated large direct and indirect costs, including healthcare and societal costs as well as costs associated with work absenteeism and loss of productivity. One study found that economic costs of ACLR and QALYs were similar at 2 years postinjury regardless of surgical technique.55 Data from the KANON trial indicate that the mean cost and QALYs associated with early ACLR do not differ from optional delayed ACLR at 5 years postsurgery.54 Additionally, analysis of the COMPARE RCT reported higher healthcare system costs and societal costs following early ACLR compared with rehabilitation and optional delayed ACLR, and early ACLR was not determined to be cost effective.58 There are other studies available comparing the cost of ACLR graft types,89 90 early versus delayed ACLR91 or ACLR versus non-operative treatment.92–94 However, the decision models used in these studies were informed by literature review and a range of assumptions based on expert opinion, some of which do not align with current evidence.91 94 These studies did not meet our eligibility criteria as they were not based on real-world data (ie, theoretical patients), so could not describe sample characteristics including time from injury to follow-up. A systematic review of all cost-effectiveness studies, that includes critical appraisal of model assumptions, could provide a useful overview of the potential long-term costs of ACL injury.
Knee symptoms and structural osteoarthritis
ACL and meniscal injury is associated with a high risk of knee osteoarthritis,95 96 and sevenfold increased odds of undergoing TKR.97 We anticipated that the long-term burden of an ACL injury would be greater in individuals with knee osteoarthritis, and thus analysed this group separately. The two studies that compared outcomes in individuals post ACL injury with and without radiographic osteoarthritis found similar HRQoL44 and physical activity levels51 between groups. However, people with knee symptoms post-ACL injury report worse HRQoL than those with few or no symptoms, irrespective of radiographic osteoarthritis status.35 The discord between structural knee osteoarthritis and patient-reported symptoms has been well documented in the literature.98 Further studies are needed to better understand the personal, economic and societal burden of symptomatic osteoarthritis following ACL injury. Given the worse HRQoL in those with symptomatic osteoarthritis, developing strategies to reduce the risk of symptomatic osteoarthritis after ACL injury may improve longer-term HRQoL and reduce the rate of TKR.
Meniscal injury
The three studies that evaluated burden (HRQoL) following meniscal injury were high-RoB or susceptible-to-some bias and included a small sample of patients undergoing meniscal surgery. Findings from two studies suggest that HRQoL may be similar 2–13 years after isolated meniscal surgery and meniscal plus ligamentous surgery.40 43 Given the documented long-term impact of meniscal tear on symptomatic knee osteoarthritis,99 there is a need for high-quality studies including patients with different types of meniscal injury undergoing surgical and non-surgical treatment to determine long-term consequences.
Strengths and limitations
First, few studies were of low RoB and RoB was accounted for through sensitivity analyses. Due to the variability in reporting sample characteristics, a meta-regression could not be performed with sufficient precision to understand the variance explained by confounding (eg, concomitant injuries, age, preinjury activity level). Only three studies included participants with an isolated ACL injury, other studies had heterogeneous inclusion criteria regarding the type and severity of concomitant injuries that were eligible, and studies varied in the amount of information provided on concomitant injuries. As a result, we combined all ACL injury studies in meta-analysis irrespective of concomitant injuries, and we were unable to explore the potential influence of concomitant injuries on long-term outcomes. The protocol was revised after a preliminary search of the literature to expand the eligible follow-up time from ≥5 years postinjury, to ≥2 years postinjury. As per the protocol, additional subgroup analyses were planned if five or more studies reported data for a given subgroup. However, this condition was only met for one subgroup analysis (time since knee injury). There was high heterogeneity in the meta-analyses, which decreases the weight of these inferences and contributed to the low certainty of evidence. Furthermore, since the aim of this review was to assess outcomes after ACL or traumatic meniscal injury, studies that did not report time from injury to follow-up (or time from injury to surgery and surgery to follow-up) were excluded. As few studies were identified following traumatic meniscal injury, enabling the inclusion of meniscal surgical studies may provide useful information related to long-term burden.
Conclusions
Physical aspects of HRQoL (ie, physical function, role limitations due to physical health, pain and general health) were worse with increased time since ACL injury, and ACL-injured groups reported worse physical aspects of HRQoL compared with uninjured controls after removal of high risk of bias studies. In contrast, mental aspects of HRQoL (ie, vitality, social function, role limitations due to emotional problems and mental health) demonstrated little change overtime after ACL injury and were similar to uninjured controls. ACL injury can result in prolonged work absenteeism, modified activity participation and substantial health and societal costs. Early ACLR for acute ACL injury was less cost-effective than rehabilitation and optional delayed ACLR. The certainty of evidence was low or very low. There is a paucity of research evaluating the long-term burden of traumatic meniscal injury.
Ethics statements
Patient consent for publication
Ethics approval
Not applicable.
Acknowledgments
The authors would like to acknowledge the other OPTIKNEE review leads Bjørnar Berg, Pætur Holm, Erin Macri, Britt Elin Øiestad, May Arna Risberg, and Anouk Urhausen, for their methodological input.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Twitter @stephfilbay, @STSkou, @yegphysio, @amraisanen, @clo2me, @agculvenor, @jwhittak_physio, @ewa_roos, @CarolynAEmery
Contributors All authors conceived the study idea and designed the study; SRF, STS, GSB, CYL, AMR, CT, AME and CE contributed to screening articles and full texts, data extraction and risk of bias appraisal; KAH codesigned and ran the search; GSB and SRF analysed the data and produced figures; SRF and CBJ performed GRADE certainty of evidence assessment; SRF drafted the first version of the manuscript and subsequent revisions; all authors contributed to editing of the manuscript and approved the final version.
Funding This review is part of the OPTIKNEE consensus (https://bit.ly/OPTIKNEE) which has received funding from the Canadian Institutes of Health Research (OPTIKNEE principal investigator JLW #161821). SRF and AGC are supported by a National Health and Medical Research Council (NHMRC) Investigator Grant (#1194428 and #2008523, respectively), STS is supported by a programme grant from Region Zealand (Exercise First) and two grants from the European Union’s Horizon 2020 research and innovation programme, one from the European Research Council (MOBILIZE, grant agreement No 801790) and the other under grant agreement No 945377 (ESCAPE). CYL is supported by The Arthritis Society Training Graduate PhD Salary Award (TGP-19-0400) and Canadian MSK Rehab Research Network 2017 Trainee Award (CIHR FRN: CFI-148081). CT is supported by a Health Research Board (HRB) Emerging Investigator Award (EIA-2019-008). AME is supported by the Canadian Institute for Health Research. JLW is supported by the Michael Smith Foundation for Health Research Scholar Award (SCH-2020-0403) and the Arthritis Society STAR Career Development Award (STAR-19-0493).
Competing interests SRF is an Associate Editor with the Journal of Science and Medicine in Sport and a board member of Osteoarthritis Research Society International (OARSI), JLW and AGC are Associate Editors of the British Journal of Sports Medicine (BJSM). JLW is an Editor with the Journal of Orthopaedic & Sports Physical Therapy. KMC is a senior advisor of BJSM and holds a research grant from Levin Health outside the submitted work. EMR is Deputy Editor and CBJ an Associate Editor of Osteoarthritis and Cartilage. EMR and STS are cofounders of Good Life with Osteoarthritis in Denmark (GLA:D®), a not-for-profit initiative hosted at University of Southern Denmark aimed at implementing clinical guidelines for osteoarthritis in clinical practice. STS is also associate editor of the Journal of Orthopaedic & Sports Physical Therapy, has received grants from The Lundbeck Foundation and personal fees from Munksgaard, all of which are outside the submitted work. All other authors declare no competing interests.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.