Arthroscopy to treat osteoarthritis of the knee?

Arthroscopy is commonly used to treat people with knee pain due to osteoarthritis. However, data collected from randomised controlled trials over the past two decades have provided compelling evidence that, in general, arthroscopic treatment comprising debridement and lavage is no more effective than placebo surgery or non-operative alternatives,1,2 and observational data suggest that it may mean a joint replacement is needed sooner.3

In this issue of the Journal, Bohensky and colleagues report a decrease in the overall rate of elective knee arthroscopy, when figures were adjusted for all elective orthopaedic procedures performed in Victoria between 2000 and 2009, but report no significant decline over time in arthroscopy rates for knee osteoarthritis.4 It is difficult to confirm the validity of the findings, due to the absence of raw data (such as absolute numbers of arthroscopies performed, by year and procedure code), the known exponential growth of elective orthopaedic procedures (which may have overestimated the decline in knee arthroscopy), unknown numbers of orthopaedic surgeons, and the inclusion of some procedural codes such as ligament reconstruction and patellofemoral stabilisation (which are not related to osteoarthritis and may have biased the results). But these observations raise pertinent questions about what is needed in order for evidence to successfully influence practice.

It is well accepted that rigorous evidence from appropriately designed and executed research studies should inform clinical practice, but there are many factors influencing evidence uptake, particularly for treatments that are already well entrenched as standard of care. These include:

• clinician factors, such as anecdotal experience, training, knowledge, and competing interests (eg, financial incentives and professional status)

• patient expectations of care and preferences

• expectations of the referring doctor

• system factors, such as access to care, health insurance, funding policy, and the ability to modify policy in response to changing evidence.

Bohensky and colleagues showed that most arthroscopies occurred in private hospitals. Further information is required to interpret these findings, as there are many factors that might influence private versus public hospital arthroscopy rates, including numbers of surgeons in each setting and whether these have changed over time, and different access to care, among other factors.

In contrast to new drugs, promising novel surgical interventions continue to be introduced into practice before their proper evaluation. However, without rigorous evidence, we cannot know whether patient outcomes are truly improved. Furthermore, once procedures are established, conducting randomised placebo-controlled trials becomes arduous, and if the results run contrary to popular belief, they are likely to be dismissed.

The negative results of the first placebo-controlled trial of arthroscopy for knee osteoarthritis by Moseley and colleagues,5 published more than 10 years ago, provoked two major responses. One response, illustrated by the editorial that accompanied the trial publication, was to praise the rigorous methods used, and to conclude that lavage and debridement were probably not efficacious for knee osteoarthritis, although it might still be appropriate for some indications. The other response, exemplified by some of the letters to the editor, concluded that as the results conflicted with personal experience, the results were erroneous and likely to be explained on the basis of methodological flaws.

Subsequent studies have verified the negative results of the Moseley trial, and criticisms regarding age, sex, symptomatology, disease severity, outcome scores and presence of malalignment, mechanical symptoms and meniscus tears have all been addressed,6,7 but it has remained difficult to shift the convictions of many surgeons. This may be due to cognitive dissonance: the need to refute or reject information that is inconsistent with ingrained beliefs.8

Other studies that have investigated therapeutic arthroscopy use for knee osteoarthritis have reported mixed results.3,9,10 Comparison between studies is made difficult due to differences in datasets, diagnostic and procedural codes, and a lack of coding verification from clinical records. Interpretation of changes over time is also hampered by the marked increase in use of magnetic resonance imaging to investigate knee pain, which, while resulting in a decline in use of diagnostic arthroscopy, has resulted in a concomitant increase in therapeutic arthroscopy. The availability of alternative non-surgical and surgical treatments (eg, viscosupplementation, glucosamine, osteotomy and partial or total knee replacements), variations in access to care, the affordability of private health insurance, the number of trained arthroscopists, and referral patterns are also important considerations in study comparisons.

There have been profound advances in trial methodology and evidence synthesis, and increased recognition of the importance of evidence to inform practice and policy, but the changing of clinician beliefs and behaviour, even in the face of credible evidence, remains highly challenging. Although there are numerous barriers that need to be overcome, improving clinician understanding of evidence is of particular importance.

Action is needed to change the current situation of making novel but untested non-drug treatments available to patients before their rigorous evaluation. The use of arthroscopy for knee osteoarthritis has been allowed to continue, exposing patients to an intervention that is at best ineffective, and at worst, harmful.