News from Nov 17, 2008

There was another article in the New England Journal of Medicine pointing out the relative ineffectiveness of arthroscopy for arthritis of the knee. (They'll keep running them every couple of years til the word gets out.) Writing in the same issue, an esteemed colleague offered some insightful observations about the study.

One of his comments, though, is hard to support: namely, "...the lack of efficacy of arthroscopic surgery in this trial does not imply that it has no role in the treatment of patients who may have osteoarthritis and also another knee condition, such as a symptomatic meniscal tear".

The study under discussion certainly did not disprove the effectiveness of arthroscopy for meniscal tears in the setting of arthritis. In Popperian terms, the null hypothesis "Arthroscopy for meniscal tears in the setting of arthritis is helpful" was not falsified. But that's the wrong null hypothesis to consider.

A null hypothesis represents what's assumed to be true until there's evidence to reject it. It's hardly appropriate to start from the premise that therapies are efficacious until proven otherwise.

The obvious effectiveness of so many orthopedic procedures - the internal fixation of femur fractures, say —should not lull us into a state of intellectual complacency in which all our treatments, by default, are assumed to work. It's just not so.