Article Text

ACL and meniscal injuries increase the risk of primary total knee replacement for osteoarthritis: a matched case–control study using the Clinical Practice Research Datalink (CPRD)
1. Tanvir Khan1,
2. Abtin Alvand2,
3. Daniel Prieto-Alhambra2,3,
4. David J Culliford4,
5. Andrew Judge2,3,
6. William F Jackson2,
7. Brigitte E Scammell1,
8. Nigel K Arden2,3,
9. Andrew James Price2
1. 1 Academic Orthopaedics, Trauma and Sports Medicine, Division of Rheumatology, Orthopaedics and Dermatology, School of Medicine, University of Nottingham, Nottingham, UK
2. 2 NIHR Musculoskeletal Biomedical Research Unit, Nuffield Department of Orthopaedics, Rheumatology and Musculoskeletal Sciences, University of Oxford, Oxford, UK
3. 3 MRC Lifecourse Epidemiology Unit, University of Southampton, Southampton General Hospital, Southampton, UK
4. 4 NIHR Collaboration for Leadership in Applied Health Research (CLAHRC) Wessex, Faculty of Health Sciences, University of Southampton, Southampton, UK
1. Correspondence to Mr. Tanvir Khan, Academic Orthopaedics, Trauma & Sports Medicine, Division of Rheumatology, Orthopaedics and Dermatology, School of Medicine, University of Nottingham, Nottingham NG72UH, UK; tanvir.khan{at}nottingham.ac.uk

## Abstract

Objectives The aim of this study was to investigate whether ACL injury (ACLi) or meniscal injury increases the risk of end-stage osteoarthritis (OA) resulting in total knee replacement (TKR).

Methods A matched case–control study of all TKRs performed in the UK between January 1990 and July 2011 and recorded in the Clinical Practice Research Datalink (CPRD) was undertaken. The CPRD contains longitudinal data on approximately 3.6 million patients. Two controls were selected for each case of TKR, matched on age, sex and general practitioner location as a proxy for socioeconomic status. Individuals with inflammatory arthritis were excluded. The odds of having TKR for individuals with a CPRD-recorded ACLi were compared with those without ACLi using conditional logistic regression, after adjustment for body mass index, previous knee fracture and meniscal injury. The adjusted odds of TKR in individuals with a recorded meniscal injury compared with those without were calculated.

Results After exclusion of individuals with inflammatory arthritis, there were 49 723 in the case group and 104 353 controls. 153 (0.31%) cases had a history of ACLi compared with 41 (0.04%) controls. The adjusted OR of TKR after ACLi was 6.96 (95% CI 4.73 to 10.31). 4217 (8.48%) individuals in the TKR group had a recorded meniscal injury compared with 669 (0.64%) controls. The adjusted OR of TKR after meniscal injury was 15.24 (95% CI 13.88 to 16.69).

Conclusion This study demonstrates that ACLi is associated with a sevenfold increased odds of TKR resulting from OA. Meniscal injury is associated with a 15-fold increase odds of TKR for OA.

• acl
• meniscus
• knee injuries
• osteoarthritis

## Introduction

The ACL and the menisci are key functional structures in the knee and both are frequently injured. Meniscal tears are the most common knee injury, with an estimated prevalence between 2.5 and 4 times higher than ACL rupture.1–4 The estimated annual incidence of ACL injury (ACLi) has been reported as high as 0.8 per 1000 population.5 The menisci are key structures in the distribution of load across the joint, and the ACL is one of the principal stabilising ligaments that couple tibial and femoral movement.6–8

There is convincing evidence that both injuries are associated with an increased risk of developing knee osteoarthritis (OA) at long-term follow-up. ACLi predisposes the knee to radiographic OA at a mean follow-up of over 10 years when comparing injured knees with uninjured contralateral knees.9 10 Meniscal injury and meniscectomy lead to an increase in peak compressive and shear stresses on the articular cartilage.11 In the longer term, Lohmander et al 2 reported that approximately 50% of individuals who had undergone meniscectomy 10–20 years previously demonstrated signs of radiographic OA.

While the risk of developing OA has been established, there is limited evidence that ACL or meniscal injury is associated with progression to end-stage knee OA and the eventual need for joint replacement. Few studies reporting the functional and symptomatic outcomes of ACLi include more than 10 years of follow-up data.12 In a similar manner the quantitative effect of meniscal injury on the risk of end-stage OA is not known.

Painful end-stage OA is often effectively treated by total knee replacement (TKR),13 with more than 670 000 performed in the USA alone in 2012.14 As pain and extent of radiographic OA are key variables in the decision to perform a TKR,15–17 joint replacement acts as a useful surrogate for end-stage OA and has been used in pharmacological studies as a marker of severe osteoarthritic disease.18 19 Within the UK Clinical Practice Research Datalink (CPRD), ACLi, meniscal injury and joint replacement are all coded, which allows their interrelationship to be investigated.

The purpose of this case–control study was to compare and quantify the risk of undergoing a TKR for knee OA in individuals with a history of ACL rupture or meniscal injury with the risk for individuals without a prior ACL rupture or meniscal injury.

## Materials and methods

### Study design and data source

A matched case–control study of all primary TKRs performed between January 1991 and July 2011 and recorded in the CPRD was undertaken. The CPRD is a large computerised primary care database containing longitudinal data on approximately 3.6 million patients (roughly 6% of the UK population) registered at 480 general practices in the UK. It is generally considered that individuals recorded in the CPRD are representative of the wider UK population in terms of age, sex and socioeconomic status.20 General practices record patient demographics, consultations, diagnoses, specialist referrals, hospital treatment, prescriptions and test results. ‘Read codes’ are used to enter clinical information, which are standard clinical terminologies used within the UK primary care. The data quality is regulated by the Medicines and Healthcare Products Regulatory Authority.

### Participants

All patients in the CPRD with a diagnosis code for TKR from January 1990 until July 2011 were identified. Previously validated Read codes were used to identify primary TKRs.20–22 Individuals with a code for primary TKR were included in the analysis if they were aged 18 years and older at the time of TKR, and those with a recorded diagnosis of inflammatory arthritis were excluded from the study.

Each case of primary TKR recorded in the CPRD (case group) and fulfilling the inclusion criteria was matched to two controls (control group). The criteria for matching cases to controls were on the basis of age, sex and General Practice, which served to control for the confounding effects related to location of residence, such as access to healthcare and affluence of the area. Controls were selected by the CPRD knowledge centre, based on incidence density random sampling from the individuals available in CPRD fulfilling the matching criteria described, for each case subject.

The CPRD medical dictionary was used to search for Read codes for ACLi. Read codes that referred to ACL and meniscal injuries were selected by the first author and these were verified by coinvestigators (two orthopaedic surgeons and one statistician).

### Data management and statistical analysis

Data management and statistical analysis were performed using STATA SE V.12. Demographics of the case and control groups were compared using χ2 test for categorical variables and t-test for continuous variables. The odds of having a TKR for individuals with ACLi recorded as an event were compared with those without ACLi using conditional logistic regression. The results were adjusted for body mass index (BMI), previous fractures around the knee and history of meniscal injury. BMI data were missing for 25 282 (16.4%) individuals. Multiple imputation methods were used in order to account for missing BMI information.23 24

The conditional logistic regression model was also used to determine the unadjusted and adjusted odds (adjusted for previous fractures around the knee, a history of ACLi and BMI) of undergoing TKR for individuals with meniscal injury recorded as an event compared with those without a documented meniscal injury.

To establish the odds of TKR for individuals with both a recorded ACLi and a meniscal injury compared with those with a history of ACLi only, an interaction term was added to the conditional logistic regression model.

We hypothesised that entry of individuals with ACLi or meniscal injury onto the CPRD might have become more robust in later years, due to an increased access to MRI scans, which may impact the comparative odds. Hence, we conducted a sensitivity analysis where TKR cases performed before 1 January 2000 and matched controls were excluded.

## Results

Within the study period, 52 530 patients in the CPRD had a primary TKR for any indication and were matched to 105 060 controls. After excluding patients with a recorded diagnosis of inflammatory arthritis, there were 49 723 individuals with TKR and 104 353 controls.

The mean age of matched individuals was 70.4 (SD 9.5) and 57.5% of them were women. BMI was higher for the TKR group. The clinical characteristics of the two groups are summarised in table 1.

Table 1

Characteristics of Clinical Practice Research Datalink-recorded cases of total knee replacement and matched controls

One hundred and fifty-three (0.31%) TKR cases had sustained an ACLi compared with 41 (0.04%) controls. The unadjusted OR of undergoing TKR, within the 20-year period, in individuals with a recorded ACLi compared with individuals with no history of ACLi was 8.00 (95% CI 5.61 to 11.42). Following multivariable adjustment, the adjusted OR was 6.96 (95% CI 4.73 to 10.31).

There were 4217 (8.48%) cases who had a recorded diagnosis of meniscal injury compared with 669 (0.64%) controls. The unadjusted and adjusted ORs (adjusted for BMI, ACLi and fractures around the knee) for TKR in individuals with a recorded meniscal injury versus those without were 15.31 (95% CI 13.99 to 16.75) and 15.24 (95% CI 13.88 to 16.69), respectively. Fractures around the knee and BMI, which were both strongly associated with risk of TKR, were included in the final model (table 2).

Table 2

Unadjusted and adjusted ORs for total knee replacement in case versus control groups with respect to each covariate included in the model

Table 3 summarises the characteristics of TKR patients in individuals with a previous ACLi and those without. Age at TKR was significantly lower in individuals who had a recorded ACLi (P<0.0001) compared with those without a previous ACLi. Individuals with a recorded meniscal injury were also significantly younger at time of TKR than those without a previous meniscal injury (P<0.0001).

Table 3

Characteristics of all Clinical Practice Research Datalink-recorded patients with total knee replacement by prior diagnosis of ACL and meniscal injury

Thirty-nine (0.07%) individuals in the case group had a recorded diagnosis of both ACLi and meniscal injury compared with three (0.003%) individuals in the control group (P<0.001). The adjusted OR for TKR in individuals with both a recorded ACL and meniscal injury compared with those with only an ACLi recorded was 4.19 (95% CI 1.05 to 16.66).

In a sensitivity analysis excluding patients who underwent TKR before 1 January 2000, there were 42 722 cases and 88 929 controls available for analysis. The unadjusted OR of TKR in individuals with an ACLi versus those without was 7.74 (95% CI 5.42 to 11.06). After adjusting for BMI, previous fractures around the knee and a history of meniscal injury, the OR was 6.81 (95% CI 4.59 to 10.11). For meniscal injury, the unadjusted odds of TKR was 15.46 (95% CI 14.06 to 17.00). After adjustment for BMI, history of ACLi and previous fractures around the knee, the OR was 15.38 (95% CI 13.96 to 16.96)

## Discussion

This matched case–control study demonstrates that ACL and meniscal injuries are both significant independent risk factors for end-stage OA treated by TKR, with an estimated 7-fold and 15-fold increased odds, respectively. This is the first epidemiological study to quantify this important link based on 20-year longitudinal data. Individuals who underwent TKR after ACL or meniscal injury were found to be significantly younger at time of TKR than those without a previously recorded diagnosis of these injuries.

A number of previous studies have demonstrated a high incidence of radiographic OA after ACLi.25–29 From relatively small cohort studies, it is difficult to quantify the influence of ACL rupture on the long-term risk of developing OA without a control group of uninjured patients. Ajuied and colleagues,9 in a meta-analysis, reported a relative risk of 3.84 for developing moderate or severe radiological OA (Kellgren and Lawrence grade III or IV) at a mean of 10 years of follow-up in ACL-injured individuals compared with those with no history of ACLi. Due to the case–control design of the present study, relative risk cannot be calculated directly, but the ORs can be used as a measure of the strength of the association between exposure and outcome. In comparison with Ajuied and colleagues’s meta-analysis, although similar, the overall probability of OA appears higher in the present study. There are several possible explanations for this. First, the follow-up period was substantially longer in the present study, which is likely to contribute to the difference. Second, the meta-analysis assessed studies comparing radiographic changes in the ACL injured with the contralateral uninjured knee within individuals. In contrast, the present study compared odds in individuals with a recorded diagnosis of ACLi with those without. The present study is therefore likely to be a closer estimate of the probability of end-stage OA attributable to ACLi within the population.

Studies reporting the long-term outcomes of surgically treated meniscal injuries have demonstrated a similar association to the present study with an increased risk of OA.30 31 In a recent systematic review of OA after total or partial meniscectomy, Papalia et al 31 reported an overall mean prevalence of OA of 53% on the operated knee, compared with a range of 0%–44% in the contralateral non-operated knee at a mean follow-up of 13.3 years. Furthermore, early structural changes following meniscal injury and meniscectomy, which result in an increased risk of developing radiographic OA, have been reported.30 32 Roemer et al using data from the Osteoarthritis Initiative, demonstrated that the presence of MRI-detected meniscal damage 2 years prior to incident radiographic OA increased the risk of incident radiographic OA. The OR for the medial meniscus was 1.83 (95% CI 1.17 to 2.89) and 1.56 (95% CI 0.85 to 2.84) for lateral meniscus.32 In the present study we have demonstrated a previously unreported significant association between meniscal injury and end-stage knee OA resulting in TKR. It is likely that some, but not all, of the individuals identified in the case and control groups with a prior diagnosis of meniscal injury would have received surgical treatment. The Read codes used were to identify individuals in the CPRD with an acute meniscal injury as opposed to meniscal surgery as this could introduce a potential confounder due to some patients undergoing meniscectomy having degenerate meniscal tears in association with established OA. Although we cannot comment on whether or not the method of treatment of acute meniscal injury influences the risk of developing severe OA, this study provides important insight into the natural history of this condition.

The present study has some limitations. As in any case–control study, there is the possibility of recall bias, although this is minimised in studies based on electronic medical records, where both the outcome (TKR) and the previous exposure (here ACLi and meniscal injury) were coded at the time when they happened. Nevertheless, the risk of more accurate recording of orthopaedic surgery among patients diagnosed with an ACLi or meniscal injury, as well as an increased provision of TKR for patients previously in contact with orthopaedic surgery (due to ACLi or meniscal injury), leads to similar issues (ie, Berkson’s bias33). We were limited by the recording of ACL and meniscal injuries in the CPRD, and no individual validation of exposure or event has been undertaken. However, if such diagnoses were under-recorded, this is likely to affect both the case and control groups approximately equally and therefore should not significantly affect the calculated ORs. TKR coding in CPRD has been recently validated with good accuracy when compared with hospital admission records.34 35 Details relating to timing of ACL/meniscal injury and TKR, as well as the specific nature of the injuries, would have provided useful information for subanalysis but within the case–control design of this study were outside the remit of this investigation. Another limitation is that we were not able to establish from the CPRD the laterality of neither the TKR nor ACLi for the individuals. Although this warrants some caution in interpreting the exact value of the calculated OR, it is unlikely to alter the overall conclusions as the assumption that the side of injury is the same as the side of TKR is an assumption for both cases and controls. In a population-based study of this magnitude with over 150 000 individuals, the influence of misrepresenting laterality of injury and TKR should not significantly change the overall OR between cases and controls.

The case–control design of the study did not allow us to undertake an analysis of operative/non-operative management of ACL and meniscal injuries, which is perhaps an important potential confounder. With respect to ligament reconstruction, two previous population-based cohort studies have addressed this question.36 37 A study based on records for cruciate ligament reconstruction (may include posterior cruciate ligament) procedures reported a seven times higher incidence of TKR after cruciate ligament reconstruction compared with a matched cohort of individuals from the general population.36 Although an area of some debate, if it is assumed that ACL reconstruction does not protect against the risk of TKR in ACL-injured patients,37 the results of the present study are not dissimilar to those reported by Leroux et al.36 However, due to the difference in study design, direct comparisons are not possible. For meniscal surgery, Lohmander et al reviewed 41 studies and reported an overall estimated OR of 10 for radiographic knee OA 15–20 years after meniscectomy when compared with an age-matched and sex-matched group with no history of knee injury.2 The authors did indicate in their review that the better designed cohort studies reported a much higher incidence of radiographic OA after meniscectomy, which would equate to a higher OR compared with age-matched and sex-matched uninjured individuals.

It has been estimated that the lifetime risk of TKR for an individual aged 50 is as high as 10.8% for women and 8.1% for men.20 Since ACL and meniscal injuries are increasing in incidence,38 39 and given the findings of this study, it is reasonable to assume that a significant number of TKRs performed are the sequelae of these injuries. It follows that the socioeconomic impact of ACL and meniscal injuries is substantial in both the short term and long term. Mather et al 40 estimated the lifetime burden of ACLi in the USA to be between $7.6 billion and$17.7 billion depending on treatment strategy. Based on the US model, the same study suggested 25 000–30 000 TKRs per year are attributable to ACLi.40 The present study by indicating a strong association between ACLi and TKR supports the claim that ACLi contributes to the financial burden of end-stage OA resulting in knee arthroplasty. Therefore, strategies to prevent knee injury including targeted training programmes41–43 as well treatment methods to avoid the onset and progression of OA following these injuries are likely to reduce the TKR burden.

## Conclusions

In conclusion, this study demonstrates a strong association between ACL and meniscal injuries and the risk of TKR. Further work should focus on determining whether initial treatment of the meniscal and ACLi has any effect in reducing the long-term risk of end-stage knee OA.

### What are the findings?

• ACL injury is associated with an approximately sevenfold increased odds of total knee replacement for osteoarthritis.

• Meniscal injury is associated with an approximately 15-fold increased odds of total knee replacement for osteoarthritis.

### How might it impact on clinical practice in the future?

• The ability to provide patients with a quantifiable risk of end-stage osteoarthritis after ACL and meniscal injuries.

• To compare long-term results of ACL and meniscal injury treatments with a baseline to ascertain if surgical therapies lower or increase the risk of end-stage osteoarthritis.

None

## Footnotes

• Contributors All authors listed fulfil the ICMJE recommendations (2013) for authorship based on four criteria: substantial contributions to the conception or design of the work, or the acquisition, analysis or interpretation of data for the work; drafting the work or revising it critically for important intellectual content; final approval of the version to be published; and agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

• Funding The study was funded by ARUK Centre for Sport, Exercise and Osteoarthritis and NIHR Academic Clinical Fellowship Award.

• Competing interests None declared.

• Ethics approval Approval was obtained prior to commencement of this study from the Independent Scientific Advisory Committee of the Clinical Practice Research Datalink.

• Provenance and peer review Not commissioned; externally peer reviewed.

## Request Permissions

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.