You might think that with the hundreds of billions spent on knee replacement since the 1970s that we would have many high-level studies showing it works. You would be wrong. In fact, the world’s first randomized controlled trial of knee replacement versus physical therapy was just published, and while it looked superficially good for the invasive procedure, a closer look shows that fewer patients should be getting knee replacements.
How Did the Study Work?
The authors took 100 knee arthritis patients and randomized them to get either physical therapy or knee replacement to determine the robustness of knee replacement results. The patients were followed using a functional questionnaire called the KOOS that measured how much they could do. The patients who had physical therapy were also allowed to get a knee replacement by the end of the first year.
What Did It Show?
At first blush, the functional questionnaire was better for the knee replacement group. The difference was statistically different, so end of story, right? Not so fast. There’s a little problem called clinically meaningful data.
How Do We Decide What’s Clinically Meaningful?
Did knee replacement results really beat physical therapy in this study? Not really, but the reasons why are a bit hidden. Let me explain.
There are two problems here—the MCID and NTT. What do these acronyms mean?
- MCID—minimal clinically important difference
- NTT—number needed to treat
The MCID Problem
The MCID is the least amount of change that a questionnaire, like the KOOS, can record that still represents something meaningful to a real patient. So it’s interesting that the authors of this paper fail to mention it. For the KOOS it’s 10 points. So that means that any change below 10 points is not really meaningful. Here the change due to knee replacement is just slightly above 10 points, hence the real clinically important difference between knee replacement and physical therapy results that we can count on is very small.
The NTT Issue
If the MCID problem shows cracks in the knee replacement wall, the NTT is the dam breaking. It’s often the single most clinically important number that we have to quickly decide if a therapy is worth its salt.
The number needed to treat is what it sounds like: the number of patients that need to be treated to get one patient to respond. If it’s a 1, it means that every patient will respond. If it’s a 10, that means that 10 patients need to be treated to get 1 to respond.
In this case the authors determined that the response they would use to calculate the NTT was a lousy 15% improvement. When they did this, the NTT was five–six!! That means that five–six patients needed to have their knees amputated to get just one who would report improvement! That fact should have been front and center in the paper, given that knee replacement surgery carries huge surgical risks as already discussed many times on this blog.
The “At the End of the Day” Measurement That Also Should Have Been Featured
There’s another hidden metric in this paper that was downplayed. The authors included what I call an “At the End of the Day” measurement. This means that this measurement was a way to quantify how good the results were in the physical therapy group. To do this, they let all patients randomized to physical therapy know that they could get a knee replacement at any time if they were unsatisfied with their results. By the end of one year, how many patients took them up on the offer? You would think from the way the knee replacement device companies advertise smiling older people climbing mountains, riding bikes, and playing tennis that all of the PT group would decide they wanted a shiny artificial knee. In reality, three in four patients who were in the PT group said “no thanks” and thought their results were good enough. Let that sink in a moment. The vast majority of the knee arthritis patients who just had physical therapy weren’t interested in getting a knee replacement. Ouch!
What These Lackluster Results Mean
Given that the results are not great compared to the risk one undertakes with this invasive procedure, to me this means that the days are numbered for knee replacement surgeries. I would venture an educated guess that stem cell injections will eventually replace this procedure—as they’ll be more cost effective with much less risk to the patient.
The upshot? This new study shows that knee replacement results are, on balance, poor given that the procedure represents the surgical amputation of a joint, which is always a risky proposition. Why wasn’t the lackluster NTT featured in the article? Knee replacements are a multibillion-dollar medical behemoth. My guess is that the authors didn’t want to piss off the sleeping giant.