I have previously written that knee arthroscopy for osteoarthritis is no more effective than alternatives, including placebo. One criticism of those studies was that arthroscopy is usually done for a torn meniscus (often incorrectly called a torn cartilage) rather than arthritis, despite the facts that the original sham trial of arthroscopy included patients with meniscus tears, and later comparative studies looked specifically at patients with torn menisci. Now, however, we have evidence from a placebo-controlled trial of arthroscopic surgery performed specifically for a torn meniscus in patients without arthritis. Evidence that shows that while most patients improve after surgery, they improve equally well after placebo surgery.
The study was reported in the New England Journal of Medicine and included 146 patients with persistent knee symptoms consistent with a meniscus tear, a proven meniscus tear on MRI and no significant arthritis, who were randomised to either sham surgery or surgery to remove the torn parts of the meniscus (usual practice). Allocation was concealed, follow up was excellent, and blinding was extended to patients, researchers, assessors, and statisticians; even to the extent that the paper was written prior to unblinding.
The results showed that for all of the primary outcomes, there was no significant (or clinically important) difference between the groups (see image for a visual of the main outcome over 12 months). This was a high quality study, it is consistent with previous studies (here and here), and it showed very little difference between the groups for multiple outcomes over all time points, so the likelihood that there is any difference between real surgery and sham surgery is low.
Having established the validity of the study results, we now face the question of generalisability: do the results apply to you, or to other groups of patients? The study found no difference in the results if the patients had acute onset of symptoms or not, but excluded patients with traumatic onset. The results are therefore not generalisable to patients with acutely torn menisci or locked knees. The results are also not generalisable to patients with lateral meniscus tears, as it was restricted to tears of the medial meniscus. However, arthroscopies are most commonly done for patients with medial knee pain who have a torn meniscus on MRI, and/or a clinical picture consistent with a torn meniscus – exactly the type of patients that were included in this study.
“But patients seem to get better”
The other interesting thing about this study was that most patients got better anyway. This is very important because it harks back to one of my themes: people often improve after treatment, so the best way to reliably attribute that improvement to the treatment is to compare it against placebo treatment in a rigorous scientific test. If it fails, like arthroscopy has repeatedly, then the improvement seen was not directly due to the treatment.
The usual question that follows the explanation above is: who cares – if patients get better, why not keep doing the treatment? My answer to that has not changed much over time: 1) because of the financial cost (which I am not happy to bear); 2) because of the opportunity cost; 3) because it carries potential harm without specific benefit; and 4) because it removes the main separation between mainstream (science-based) medicine and alternative medicine (which survives on indirect and placebo effects).
The bottom line
If you have several months of knee pain that is consistent with a medial meniscus tear, and if you have a meniscus tear on MRI, your symptoms are likely to improve. Having the torn parts of the meniscus removed will not change that in any meaningful way.