Back pain is a huge problem: it affects many people, causes much suffering and leads to extraordinary high cost for all of us. Considering these facts, it would be excellent to identify a treatment that truly works. However, at present, we do not have found the ideal therapy; instead we have dozens of different approaches that are similarly effective/ineffective. Two of these therapies are massage and acupuncture.
Is one better than the other?
This study compared the efficacy of classical massage (KMT, n = 66) with acupuncture therapy (AKU, n = 66) in patients with chronic back pain. The primary endpoint was the non-inferiority of classical massage compared with the acupuncture treatment in respect of the impairment in everyday life, with the help of the Hannover function questionnaire (HFAQ) and the reduction in pain (“Von Korff”-Questionnaire) at the follow-up after one month.
In the per-protocol analysis during the period between enrollment in the study and follow-up, the responder rate of the KMT was 56.5% and thus tended to be inferior to the responder rate of the AKU with 62.5% (Δ = - 6%; KIΔ: - 23.5 to + 11.4%).
The authors concluded that classical massage therapy is not significantly inferior to acupuncture therapy in the period from admission to follow-up. Thus, the non-inferiority of the KMT to the AKU cannot be proven in the context of the defined irrelevance area.
I find such studies oddly useless.
To conduct a controlled trial, one needs an experimental treatment (the therapy that is not understood) and compare it with an intervention that is understood (such as a placebo that has no specific effects, or a treatment that has been shown to work). In comparative studies like the one above, one compares one unknown with another unknown. I do not see how such a comparison can ever produce a meaningful result.
In a way, it is like an equation with two unknowns: x + 5 = y. It is simply not possible to define either x nor y, and the equation is unsolvable.
For comparative studies of two back-pain treatments to make sense, we would need one of which the effect size is well-established. I do not think that we currently have identified such a therapy. Certainly, we cannot say that we know it for massage or acupuncture.
In other words, comparative studies like the one above are a waste of resources that cannot possibly make a meaningful contribution to our knowledge.
To put it even more bluntly: we ought to stop such pseudo-research.
I agree with you on the futility of such a trial – even if they were able to demonstrate a difference this would be compatible with one treatment making the patient worse (or indeed both of them).
However, if non-inferiority was the primary endpoint in the trial design, then they should have taken advice from a statistician as to the sample size required. Indeed there are easy-to-use online tools available for this calculation. Putting approximate figures into one of them, assuming that we want our trial to have a 90% chance of finding a 10% difference at the 95% significance level where the response rate is expected to be about 60%, we get a required sample size of 324, which is roughly three times the size of this trial.
In other words, it appears that, whatever the authors’ intentions, the trial was designed from the start not to show anything at all, and the statistical analysis they used was entirely spurious.
In my opinion, it depends on the person itself which one they will choose to remove the back pain in their system, furthermore, either one of them can be very helpful.