rss
Br J Sports Med 47:66-70 doi:10.1136/bjsports-2012-091490
  • Review

Is patellofemoral joint osteoarthritis an under-recognised outcome of anterior cruciate ligament reconstruction? A narrative literature review

  1. Kay M Crossley1,3
  1. 1Division of Physiotherapy, School of Health and Rehabilitation Sciences, The University of Queensland, Brisbane, Queensland, Australia
  2. 2Department of Physiotherapy, School of Primary Health Care, Monash University, Melbourne, Victoria, Australia
  3. 3Department of Mechanical Engineering, The University of Melbourne, Parkville, Australia
  4. 4Department of Physiotherapy, Melbourne School of Health Sciences, The University of Melbourne, Parkville, Australia
  1. Correspondence to Dr Kay M Crossley, Division of Physiotherapy, School of Health and Rehabilitation Sciences, The University of Queensland, Building 84A, St Lucia, Queensland 4072, Australia; k.crossley{at}uq.edu.au
  • Received 18 June 2012
  • Accepted 3 September 2012
  • Published Online First 4 October 2012

Abstract

Patellofemoral joint (PFJ) osteoarthritis (OA) is a prevalent disease capable of being a potent source of knee symptoms. Although anterior cruciate ligament (ACL) injury and reconstruction (ACLR) are well-established risk factors for the development of tibiofemoral joint OA, PFJ OA after ACL reconstruction has gone largely unrecognised. This is despite the high prevalence of anterior knee pain after ACLR, which can reduce the capacity for physical activity and quality of life. The susceptibility of the PFJ to degenerative change after ACLR may have implications for current rehabilitation strategies. This review summarises the evidence describing the prevalence of PFJ OA after ACLR and examines why this compartment may be at increased risk of early onset OA after ACLR. Strategies that address the modifiable factors for risk of PFJ OA may aid in alleviating joint loads and symptoms for people after ACLR.

Introduction

The patellofemoral joint (PFJ) is increasingly recognised as an important contributor to knee joint osteoarthritis (OA).1 ,2 Indeed, radiographic PFJ OA may be more common than tibiofemoral joint (TFJ) OA in community-based studies,2 ,3 and more strongly associated with knee symptoms than TFJ OA.4 ,5 Anterior cruciate ligament (ACL) injury and reconstruction (ACLR) are well-established risk factors for TFJ OA, with degenerative radiographic changes in 10–90% of individuals 10–20 years after the injury.6 It appears that the PFJ may also be vulnerable to OA development following ACLR.7 ,8 Given that ACLR surgery is typically performed on people younger than 30 years, early onset of OA may occur some 15 years sooner than expected in the general population.9 ,10 Owing to limited conservative treatment strategies for OA, early onset OA after ACLR ultimately increases the likelihood of joint replacement surgery at a younger age. Thus, given its potential impact on physical function11 and quality of life,12 PFJ OA may represent an important consideration in those who have undergone ACLR.

Rehabilitation following ACLR focuses on returning to physical activity and sport,13 with surgical success often measured by a timely ability to achieve this goal.14. However, if the PFJ is at risk of developing OA after ACLR, then rehabilitation may also need to focus on reducing the risk factors for PFJ disease and symptoms. This narrative review aims to: (1) describe the prevalence of PFJ OA after ACLR; (2) evaluate why the PFJ may be predisposed to degenerative change after ACLR and (3) enhance understanding of the PFJ after ACLR and inform appropriate rehabilitation strategies.

Prevalence of PFJ OA after ACLR

Radiographic evaluation of PFJ OA after ACLR

Patellofemoral joint OA is characterised by loss of articular cartilage, osteophyte formation, subchondral bone change and synovitis affecting the patella or femoral trochlea groove. In contrast to more than 100 studies that have been published on TFJ OA after ACL injury and reconstruction,6 only 17 studies have reported radiographic PFJ OA after ACLR, with a prevalence ranging from 11% to 90% (median 36%), 2–15 years after surgery.7 ,8 ,15–29 Most radiographic PFJ OA after ACLR was classified as mild (84%), although moderate (15%) and severe (1%) PFJ OA was also observed (table 1). While few studies evaluated PFJ OA less than 5 years after ACLR, it appears that the prevalence increases with time since surgery. Furthermore, in those studies that described OA prevalence in both knee compartments, the PFJ and TFJ appear to be similarly affected by radiographic OA (figure 1). Notably, the median prevalence of PFJ OA 10–15 years after ACLR is almost 50%, and the median age of all individuals studied is 38 years. The prevalence and variability of PFJ OA after ACL injury and reconstruction appears similar to TFJ OA (10–90%, 10–20 years after ACL injury)6 and they frequently coexist,8 ,18 as occurs in the general population.2

Table 1

Summary of studies that report prevalence of PFJ OA after ACLR using an established radiographic classification system

Figure 1

Prevalence of patellofemoral joint and tibiofemoral joint osteoarthritis after anterior cruciate ligament reconstruction. Median and IQR of only studies that used an established classification system to assess PFJ OA. Tibiofemoral OA reported as medial and lateral compartment combined or in studies that report compartmental distribution of TFJ OA alone, the compartment with the highest OA prevalence is included. ACLR, anterior cruciate ligament reconstruction; PFJ OA, patellofemoral joint osteoarthritis; TFJ OA, tibiofemoral joint osteoarthritis.

The large variability in PFJ OA prevalence may reflect the range of radiographic diagnostic criteria used (table 1). The Kellgren and Lawrence (KL) classification system30 considers both osteophytes and joint space narrowing (JSN), whereas the International Knee Documentation Committee (IKDC)31 and Fairbank classification systems32 emphasise JSN for defining OA. The Osteoarthritis Research Society International atlas33 scores JSN and marginal osteophytes separately in the lateral and medial PFJ compartments. Despite commonly being used to define PFJ OA, the original KL-grading criteria was described for the TFJ alone30 and thus may not be appropriate to define PFJ OA.

Studies in this review used a variety of scoring systems, and three studies did not use a recognised classification method to define PFJ OA.16 ,24 ,25 Other variability in the studies arose from heterogeneous populations (differing on eligibility criteria such as combined injuries or preoperative radiographic OA), varying radiographic methods (different weight-bearing status and knee flexion angles) and different surgical procedures which are also likely to contribute to the variability in PFJ OA prevalence observed.

Influence of graft type on PFJ OA after ACLR

The relationship between ACLR and PFJ OA has mostly been investigated in those with a bone-patellar tendon-bone (BPTB) autograft reconstruction, which was the predominant surgical technique for many years.34 The investigation of PFJ OA after BPTB autograft reconstruction is not surprising given that this procedure harvests the patellar tendon and some bone from the patella, which are intimately related to the PFJ. This type of reconstruction has been associated with PFJ symptoms.35 ,36

Recently, hamstring tendon autograft use has increased in popularity and has minimal direct effect on the extensor mechanism. However, despite more frequent rates of PFJ OA noted after BPTB than hamstring tendon autograft (41% compared to 30%20; 30% compared to 16%22) the between-graft differences in PFJ OA prevalence were not significant20 and did not predict radiographic PFJ OA.22 Although further research is required to confirm these findings, the current literature suggests that hamstring tendon autograft may not protect against PFJ OA development after ACLR. The potential reasons for this are discussed later in this review. Thus, PFJ OA should be considered in any individual who has undergone ACLR, irrespective of graft type.

Arthroscopic and MRI evaluation of PFJ OA after ACLR

Although radiography remains the most accessible and utilised tool to assess OA, it is unable to detect early-stage joint disease, such as derangement of articular cartilage.37 More sensitive methods to detect changes in articular cartilage may provide valuable insight into onset and progression of PFJ OA. Detection of pre-radiographic joint changes is pertinent after ACL injury, given the likelihood that early changes may lead to longer term more severe structural joint changes and pain.6

A number of studies have utilised arthroscopy and MRI to assess PFJ articular cartilage damage, with current evidence suggesting that pre-radiographic PFJ OA is both present and progressive after ACLR. At second-look arthroscopy, approximately 18 months after ACLR, 30–57% of individuals demonstrated deterioration of PFJ cartilage since ACLR.38–41 Thus, ACLR does not appear to protect the PFJ from progression of degenerative change. Remarkably, by 7–11 years after ACL injury, irrespective of conservative or surgical management, individuals have been shown to be 30 times more likely to have patellar cartilage loss on MRI compared with baseline (time of ACL injury) (95% CI 8 to 115).42

PFJ symptoms after ACLR

Knee OA can be defined by radiographic abnormalities alone or in combination with knee symptoms.43 It is important to consider knee pain and symptoms in the diagnosis of OA, as radiographs alone do not define the clinical syndrome since many patients with radiographic changes are asymptomatic.43

Patellofemoral pain is common after ACLR, particularly with a BPTB autograft, with up to 40% of people reporting anterior knee pain during activity 10 years after ACLR.36 Indeed, patellofemoral pain is said to be one of the most serious and troublesome complications that compromises final outcome after ACLR.40 This pain is typically attributed to graft site morbidity, from removal of the middle third of the patellar tendon and associated bone, rather than PFJ degeneration. Knee flexion contracture (ie, loss of knee extension), which can increase PFJ contact forces, has also been positively correlated with patellofemoral pain.44 Although symptoms and radiographic findings are often poorly correlated in all OA,45 ,46 the PFJ has been shown to be a potent source of knee OA symptoms, more so than the TFJ.47 Even isolated mild radiographic PFJ OA can be associated with considerable symptoms that impact substantially on activities of daily living.11 ,47 Therefore, despite the majority of PFJ OA being of mild severity after ACLR, it may still have important implications for pain and function after ACLR.

There are minimal investigations regarding the relationship between radiographic PFJ OA and knee symptoms after ACLR. Jarvela et al19 found significant differences in subjective reports of pain and function on the Lysholm rating scale between those who were classified as having mild, moderate or severe PFJ OA. However, others have not found this association using alternative patient-reported measures (IKDC subjective score).23 Neuman et al7 noted a non-significant trend towards worse symptoms in individuals with radiographic PFJ OA on the Knee Injury and Osteoarthritis Outcome Score. The discrepancies that exist in the association between knee symptoms and radiographic PFJ OA may in part be explained by studies associating symptoms with either the presence of any PFJ OA,7 ,23 or with different levels of radiographic PFJ OA severity.19 Nonetheless, while further research of PFJ OA symptoms after ACLR is required, these findings highlight that anterior knee pain symptoms should be monitored post-ACLR, and the potential relationship with radiographic PFJ OA considered.

Summary of PFJ OA prevalence after ACLR

Current evidence indicates that PFJ OA is prevalent after ACLR. Although most radiographic changes are mild, PFJ OA has the potential to adversely affect pain and function after ACLR, thus highlighting the importance for clinicians to consider the PFJ during postoperative rehabilitation. PFJ articular cartilage degeneration appears to be progressive after ACLR, which likely contributes to the development of radiographic PFJ OA over time.

Proposed contributors to PFJ oa development after ACLR

Injury and reconstruction of the ACL can have adverse effects on all knee compartments; however there are events that occur during an ACL injury and after reconstruction that may have specific implications for the PFJ. Such events include concomitant damage to articular cartilage and meniscus, inflammation, biomechanical changes and quadriceps strength deficits, which may contribute to a greater risk of developing PFJ OA after ACLR.

Concomitant damage to articular cartilage and inflammation after ACL injury

The substantial force required for ACL rupture, or the instability associated with ACL deficiency, may damage articular cartilage and subchondral bone, both in the PFJ and TFJ. Moderate-to-severe cartilage lesions affecting the PFJ are observed in 2–13% of ACL injured knees,24 ,40 ,48 with mild lesions observed in 15–29%.38 ,40 ,41 ,49 ,50 Most importantly, concomitant damage to the articular cartilage, noted at the time of ACLR, predicts long-term radiographic OA in the PFJ and TFJ.21 ,22 Li et al22 reported the OR for PFJ and TFJ OA in those with articular cartilage damage in the same compartment at the time of ACLR to be 3.7 (95% CI 1.2 to 10.9) and 4.3 (95% CI 2.0 to 9.3), respectively. The time delay between injury and ACLR may also affect PFJ49 and TFJ51 cartilage detrimentally, secondary to ongoing instability. However, femoral trochlea cartilage thinning has been shown to occur 2 years after ACL injury, irrespective of conservative management or early ACLR.52

The initiation of cytokine and protease cascades in the acute phase of ACL rupture is often associated with damage to the type II collagen network, aggrecan and other components of joint cartilage,6 which may be sufficient to initiate cartilage degeneration. Inflammatory markers generally diminish with time but may be present for years after ACL injury, at similar levels to those in OA.6 Furthermore, ACLR surgery creates additional intra-articular trauma, which may prolong or initiate a further inflammatory response.38 ,53 Increased concentrations of inflammatory biomarkers occur in both serum54 and synovial fluid,55 with potential consequences for all joint compartments, including the PFJ. The articular cartilage lesions, observed at the time of ACLR, may be particularly vulnerable to progressing into more severe degenerative joint change in the presence of prolonged inflammation before and after ACLR.

Meniscal injury and meniscectomy

Meniscal and ACL injuries frequently coexist.56 Systematic reviews highlight that meniscal tears and/or surgery are established risk factors for TFJ OA after ACL injury57 and in non-ACL injured populations.58 Meniscectomy has also been shown to predict PFJ OA in those with7 ,21 and without12 ACL injury. The OR for PFJ OA 20 years after medial and lateral meniscectomy without ACL injury was 2.6 (95% CI 1.1 to 6.6) and 5.3 (95% CI 1.9 to 15.0), respectively.12 Although approximately one-third of those who went on to develop PFJ OA in this cohort had PFJ cartilage changes at the time of meniscectomy12, the development of PFJ OA in the remaining two-thirds must be attributed to alternate factors. Similarly, in those with an ACL injury, meniscectomy at the time of ACLR was associated with PFJ OA (r=0.45) at 6 year follow-up.21

An injured or resected meniscus could have mechanical effects on TFJ transverse plane rotation59 or frontal plane alignment, which may alter PFJ contact pressure.60 ,61 Although no studies have evaluated PFJ alignment after meniscectomy, altered frontal plane moments have been observed after partial meniscectomy compared with healthy controls.62 ,63 Further research is required to investigate the mechanical effects of meniscal injury and surgery, and the relationship with PFJ mechanics and subsequent PFJ morbidity. However, from the available literature, it appears that meniscus preservation, where possible after ACL injury, may protect the PFJ from OA development.

Tibiofemoral biomechanics influence patellofemoral load

Knee biomechanics may provide valuable insight into why an ACLR places the PFJ at risk of developing OA. It is well-established that TFJ biomechanics, including joint motions and loads are altered following ACL injury64 and do not appear to be restored with either a BPTB or hamstring autograft reconstruction.65–68 Altered TFJ kinematics after ACLR are hypothesised to contribute to the initiation and progression of TFJ OA by changing cartilage load distribution.69–71 Although the biomechanics of the PFJ are different to the TFJ, evidence from cadaver and modelling studies indicate that changes in TFJ mechanics can alter PFJ alignment and stress.72 ,73 Therefore, alterations in TFJ kinematics have the potential to affect PFJ mechanics and predispose to PFJ OA initiation or progression.

Injury to the ACL disrupts antero-posterior stability in the sagittal plane and rotatory stability in the transverse plane.64 Although reconstruction normally restores antero-posterior and rotatory stability, it does not consistently restore transverse plane rotation range of motion (ROM).71 Most studies concur that greater tibial rotation ROM, or ‘external rotation offset’ (decreased internal rotation/increased external rotation) is evident in the ACLR knee compared to the uninjured knee65 ,67 ,74–83 and healthy controls.67 ,68 ,77 ,80–82 ,84 This is likely to have significant implications for the PFJ, given that experimentally induced tibial external rotation is associated with increased lateral patellar tilt and rotation85 and PFJ load60 (figure 2A). Indeed, Van de Velde et al86 found that abnormal patellar rotation, tilt and lateral shift in cartilage contact occurred in vivo after ACL injury and was not restored with ACLR. These altered transverse plane TFJ and PFJ kinematics may predispose PFJ cartilage to degenerative change, theoretically preceding the development of PFJ OA.

Figure 2

Schematic drawing demonstrating how the patellofemoral joint is affected by; (A) transverse plane external tibial rotation; (B) frontal plane malalignment and (C) sagittal plane knee flexion.

Not all studies noted abnormalities in tibial rotation after ACLR.87–90 The inconsistencies in study findings may be attributed to difficulties in accurately measuring tibial rotation using three-dimensional motion analyses, and differences in task (eg, high-demand vs low-demand) or ACLR technique (eg, single-bundle or double-bundle). For example, there is preliminary evidence that a more obliquely placed femoral tunnel may correct abnormal tibial rotation more effectively,68 ,82 ,91–93 but may not be associated with PFJ OA development.94 Both the double-bundle and oblique-tunnel orientation techniques require further investigation into their ability to restore transverse plane kinematics, and thus the ability to protect the PFJ and TFJ from unaccustomed load and OA development or progression.

In the frontal plane, an increased external knee adduction moment (internal knee abduction moment) has been observed 5 years after ACLR and may present a potential mechanism for medial TFJ OA development.95 Although these results were not corroborated by different studies at 20 months96 or 6 years97 after ACLR, altered TFJ alignment in the frontal plane can affect PFJ loading patterns by increasing load on the medial or lateral patellar facet, and potentially predispose the PFJ to OA98 (figure 2B). Notably, TFJ valgus alignment may be of particular relevance for PFJ OA considering that an external knee abduction moment (internal knee adduction moment) is a risk factor for PFJ pain in adolescents.99 Interventions aimed at correcting frontal plane malalignment, such as gait retraining, hamstring and hip abductor strengthening, knee braces or medially or laterally wedged shoe inserts, which attempt to optimise TFJ load,100 may also be important for PFJ alignment and stress.

Loss of knee range of motion

In the sagittal plane, increasing TFJ flexion angles progressively increase both PFJ contact area and pressure due to the resultant quadriceps and patellar tendon forces101 (figure 2C). This is important after ACLR given that an inability to achieve full knee extension has been shown to occur in up to 21% of individuals10 ,102 ,103 and has been associated with the presence of PFJ OA7 ,19 and TFJ OA.10 ,102 Knee extension loss subjects the PFJ to a constant heightened compressive load in weight-bearing101 and is positively correlated with patellar irritability and patellofemoral symptoms.44 Loss of knee flexion after ACLR has also been associated with PFJ OA.7 It is thus imperative for early postoperative rehabilitation strategies, as recommended by Shelbourne and Nitz,104 to focus on achieving full knee extension and flexion to prevent longer term abnormal PFJ loading.

Quadriceps strength and vasti co-ordination

Muscle weakness is one of the earliest and most frequent findings in people with knee OA105 and may be a greater risk factor for PFJ OA than TFJ OA. In a rabbit model, experimental quadriceps weakness induced retropatellar cartilage degeneration, which was not seen in TFJ cartilage.106 Similarly, greater quadriceps strength protected against lateral PFJ cartilage loss in humans,107 but this relationship was not found in the TFJ.107 ,108 Since quadriceps strength deficits may persist for up to 6 years after ACLR,20 quadriceps weakness after ACLR may play a role in the development of PFJ OA. However, the only study to specifically evaluate the relationship between quadriceps strength and PFJ OA found no association,21 thus the relationship between quadriceps weakness and PFJ OA development remains unclear.

A further consideration for the relationship between quadriceps strength and PFJ OA after ACLR is the co-ordinated activation of all four quadriceps components. Optimal PFJ biomechanics relies on the interaction between the vastus medialis obliquus (VMO) and vastus lateralis.109 Inhibition or delayed onset of the VMO occurs in the presence of knee pain110 ,111 and effusion,112 ,113 which is common after ACLR.36 If alteration of the VMO function (lower magnitude and/or delayed onset timing) is present after ACLR, increased lateral PFJ stress may follow, potentially contributing to the pathogenesis of PFJ symptoms and OA. This requires further investigation in an ACLR cohort.

Summary of proposed contributing factors to PFJ OA after ACLR

Factors with potential to contribute to the development of PFJ OA after ACLR have been proposed. Inflammation and concomitant damage to the PFJ articular cartilage or meniscus at the time of, or after ACL injury and reconstruction may make the PFJ vulnerable to degeneration. The inability of an ACLR to fully restore knee biomechanics and the prolonged ROM and strength deficits that may occur even after rehabilitation may further promote derangement of the PFJ and, combined with compromised articular cartilage, lead to PFJ OA.

Recommendations and future research directions

The proposed factors involved in the development of PFJ OA require research to optimise the outcome after ACLR and minimise the burden of PFJ OA. Increasing awareness that the PFJ is an important compartment to consider after ACLR will encourage future studies to image the PFJ compartment more frequently, using appropriate imaging methods, and report specific compartmental distribution of OA.

Although it would be valuable to compare PFJ OA prevalence between surgical and conservative management after ACL injury, it was not possible in this review. The two studies that compared PFJ OA prevalence between the two management options only performed ACLR on those participants who failed conservative management, thus biasing the surgical cohort.7 ,8 Therefore, well-designed studies, ideally randomised controlled trials, are required to elucidate if conservative or surgical management has a greater impact on long-term PFJ health and symptoms.

Identifying pathophysiological and biochemical changes during the initial stages of PFJ OA development may highlight potential risk factors leading to disease onset and progression. The emergence of MRI and quantitative image analysis promise more sensitive methods for diagnosis and evaluation of articular cartilage derangement and early OA changes. Such methods include T2-weighted MRI mapping and the delayed gadolinium-enhanced MRI of cartilage (dGEMRIC) technique, which evaluate glycosaminoglycan concentration to indicate cartilage biochemical status. A number of classification systems to describe early OA changes from MRI have recently been established,114–116 and should be utilised in future studies to evaluate the PFJ after ACLR. Although MRI remains expensive in clinical practice, all routine radiographs should at least include a skyline view. In addition, clinicians should be more alert to complaints of patellofemoral pain after ACLR and more focused on preventing patellofemoral pain—a likely predictor of PFJ OA.117

While this review has not specifically investigated risk factors for ACL injury, there is emerging evidence that ACL injury shares some similar risk factors as patellofemoral pain. Hip and knee kinematics, specifically excessive hip internal rotation118 and increased external knee abduction moment (internal knee adduction moment),99 have been shown to be risk factors for patellofemoral pain and are likely to contribute to high knee valgus loads when landing from a jump.119 Furthermore, prospective measures of dynamic knee valgus during landing predict ACL injury risk.120 The link between patellofemoral pain and ACL injury may enable preventative strategies to target both problems simultaneously and reduce both ACL injuries, and hence ACLR, and post-traumatic PFJ OA. Future research could investigate whether the kinematics that drive an ACL rupture are also risk factors for the development of PFJ OA.

Another factor with potential to contribute to the development of PFJ OA after ACLR, which has rarely been considered in the literature is the impact of physical activity and return to sport.7 While measurement and reporting of physical activity is challenging, these measures should be considered in future trials. Higher patellofemoral loads may be experienced by individuals who return to sport/physical activity compared to those who do not, especially if biomechanical features that increase PFJ stress are evident after ACLR. Future studies could also investigate if either specific rehabilitation exercises have the potential to reduce PFJ stress after ACLR121 or whether there are commonly used exercises in ACLR rehabilitation that increase PFJ stress, which should be modified or avoided.

This review has proposed several theoretical mechanisms for the development and progression of PFJ OA after ACLR. However, more controlled longitudinal studies to investigate surgical, biochemical, biomechanical and musculoskeletal risk factors, associated with the early changes assessed from MRI and more advanced degeneration from radiographs, are needed. Knowledge surrounding these risk factors may enable modification of rehabilitation goals or surgical technique to address deficits associated with the development or progression of PFJ OA and symptoms after ACLR.

Conclusion

This review concludes that PFJ OA may be an under-recognised outcome of ACLR, and be at least as common as TFJ OA. The symptomatic and functional implications of PFJ OA and its influence on the short-term and long-term outcomes after ACLR are not known, despite the potential for PFJ OA to be a potent source of knee symptoms. Factors that may lead to the well-established link between TFJ OA and ACL injury and reconstruction can also affect the PFJ compartment. Articular cartilage lesions, inflammation, meniscal injury and resection, quadriceps strength deficits and altered biomechanics may all contribute to the increased risk of PFJ OA after ACLR. Optimising surgical technique, management of postinjury and postoperative inflammation, and pursuing a targeted rehabilitation may assist in preventing or ameliorating PFJ OA after ACLR.

What this study adds

  • There is a high prevalence of patellofemoral joint (PFJ) osteoarthritis (OA) after anterior cruciate ligament reconstruction (ACLR).

  • Several mechanisms exist for PFJ OA after ACLR.

  • The importance of addressing factors contributing to PFJ OA development and/or progression during ACLR rehabilitation programmes.

Footnotes

  • Contributors All authors contributed to the conception, design, writing and revisions of this review.

  • Funding The authors would like to acknowledge funding support for AC from Arthritis Victoria and an Australian Postgraduate Award, and for NC from a National Health and Medical Research Council postdoctoral fellowship.

  • Competing interests None.

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

  • ▸ References to this paper are available online at http://bjsm.bmjgroup.com

References

Free sample

This recent issue is free to all users to allow everyone the opportunity to see the full scope and typical content of BJSM.
View free sample issue >>

Email alerts

Don't forget to sign up for content alerts so you keep up to date with all the articles as they are published.