A new study tested the efficacy of chiropractic spinal manipulative therapy (CSMT) for migraine. It was designed as a three-armed, single-blinded, placebo -controlled RCT of 17 months duration including 104 migraineurs with at least one migraine attack per month. Active treatment consisted of CSMT (group 1) and the placebo was a sham push manoeuvre of the lateral edge of the scapula and/or the gluteal region (group 2). The control group continued their usual pharmacological management (group 3).
The RCT began with a one-month run-in followed by three months intervention. The outcome measures were quantified at the end of the intervention and at 3, 6 and 12 months of follow-up. The primary end-point was the number of migraine days per month. Secondary end-points were migraine duration, migraine intensity and headache index, and medicine consumption.
The results show that migraine days were significantly reduced within all three groups from baseline to post-treatment (P < 0.001). The effect continued in the CSMT and placebo groups at all follow-up time points (groups 1 and 2), whereas the control group (group 3) returned to baseline. The reduction in migraine days was not significantly different between the groups. Migraine duration and headache index were reduced significantly more in the CSMT than in group 3 towards the end of follow-up. Adverse events were few, mild and transient. Blinding was strongly sustained throughout the RCT.
The authors concluded that it is possible to conduct a manual-therapy RCT with concealed placebo. The effect of CSMT observed in our study is probably due to a placebo response.
Chiropractors often cite clinical trials which suggest that CSMT might be effective. The effects sizes are rarely impressive, and it is tempting to suspect that the outcomes are mostly due to bias. Chiropractors, of course, deny such an explanation. Yet, to me, it seems fairly obvious: trials of CSMT are not blind, and therefore the expectation of the patient is likely to have major influence on the outcome.
Because of this phenomenon (and several others, of course), sceptics are usually unconvinced of the value of chiropractic. Chiropractors often respond by claiming that blind studies of physical intervention such as CSMT are not possible. This, however, is clearly not true; there have been several trials that employed sham treatments which adequately mimic CSMT. As these frequently fail to show what chiropractors had hoped, the methodology is intensely disliked by chiropractors.
The above study is yet another trial that adequately controls for patients’ expectation, and it shows that the apparent efficacy of CSMT disappears when this source of bias is properly accounted for. To me, such findings make a lot of sense, and I suspect that most, if not all the ‘positive’ studies of CSMT would turn out to be false positive, once such residual bias is eliminated.
another comment on the same paper
What this study shows is that if you see a chiropractor, as a single chiropractor administered the chiropractic and sham adjustments, you will suffer less with migraine. I think the problem with this study was that its very difficult to administer a sham gonstead adjustment, the sham adjustment had an effect.
Funny how you didn’t highlight this bit.
“Migraine duration and headache index were reduced significantly more in the CSMT than in group 3 towards the end of follow-up.” No placebo group mentioned.
And no strokes or deaths!!!!!!!!!!
“Adverse events were few, mild and transient.”
sample size was at least one dimension too small to pick up infrequent adverse effects.
“Both real and sham [manipulation therapy] are associated in reduction in headache days relative to the usual care arm,” Lipton told MedPage Today. “Because real and sham [manipulation therapy] are equally effective, that supports one of two conclusions. One possibility is that the apparent benefits of manipulation therapy are placebo effects, induced by the laying on of hands.”
“The other possibility is that both real and sham manipulation therapy are effective treatments,”(ESPECIALLY IF DONE BY A CHIROPRACTOR) he continued. “The fact that the treatments had very favorable safety results is reassuring. Based on these results, manipulation therapy is not an evidence-based therapy. Given the benefits relative to usual care and its safety, it may be worth a try in people whose treatment needs are not met by usual, evidence-based treatments.”
2 possibilities, here is a blog without bias, unlike Prof Ernst.
a sham push manoeuvre of the lateral edge of the scapula and/or the gluteal region effective against migraine???
the possibility occurred to me too – but I concluded that it is merely theoretical.
This study as per its synopsis PROBABLY doesn’t support whatever type of manipulation was performed in the treatment of migraines; of course, we don’t know what spinal regions were treated or what manipulative techniques were used. I note that the placebo treatment was, for some reason, either a scapular or gluteal “push maneuver.” I wonder why? The synopsis didn’t state what the chiropractic treatment frequency was during the study. Therefore it’s difficult to comment on the study at this time.
One thing stands out regarding the authors’ opinion that their study represented the successful application of a concealed placebo: I doubt that even the most desperate of patients or the most voracious insult-lobbing chiro-haters on this site would believe that pushing on a shoulder blade or a buttock might cure a migraine. Placebos must be believable in order to be placebos.
“One thing stands out regarding the authors’ opinion that their study represented the successful application of a concealed placebo”
the authors checked the success of blinding.
it depends on how you inform the patients when taking informed consent.
No it’s not, just click on the link in the post. The full paper’s available free on line (a rare thing for altmed studies). If you bothered to read the paper you’d find, for example: “The CSMT group received spinal manipulative therapy using the Gonstead method, a specific contact, high-velocity, low-amplitude, short-lever spinal with no post-adjustment recoil that was directed to spinal biomechanical dysfunction (full spine approach) as diagnosed by standard chiropractic tests at each individual treatment session.”
The paper even cross-references the full trial protocol. It too is freely available. For once we don’t have to depend on a study abstract.
Thank you , Frank. I did click the link and read the synopsis but my browser wouldn’t allow me to access the full text. I will search for it on pubmed and NEXIS.
I have concerns about this study. With only 20–27 patients per group it seems grossly underpowered to provide robust data. The author’s conclusions are certainly all that can be said about the results, but even the cautious “the effects seen are probably due to the placebo response” seems to me to be on shaky ground.
The obvious question is why the control group did so badly. At first sight this should prompt a new look at “usual pharmacological management”: perhaps the painkillers used regularly for migraine control are not effective and should be abandoned. But from the paper’s table 4, we learn that the patients in the CSMT and placebo groups were also taking paracetamol, triptans paracetamol+codeine and NSAIDs at the same levels as the controls. There was just a single time-point where the average amount of each drug taken differed ‘significantly’ (p=0.03) between CSMT and controls, but that’s just one of 53 calculated p-values — clearly the one result statistically due to chance.
So, for me, the bottom line of this paper is that the study was yet another A+B vs B study, where A is any sort of extra manipulative attention and B is the normal gamut of painkillers for migraine. Proving that the placebo effect is alive and well, but we already knew that.
I think you are wrong: If by A you mean the medication – it was taken by both the SM and sham groups.
No, like I said in the comment, B is the medication, taken by all groups. A is the pooled group of CSMT + sham manipulation. A can’t lose, right?
BTW, I just realized you usually say “A+B vs. A”, but in my comment I used A to refer to the ‘added’ Rx. In future I’ll use ‘M+S vs M’ to avoid — or add to — the confusion!
the terminology is fairly well established since our paper of 2008: https://www.ncbi.nlm.nih.gov/pubmed/18626172
??! So I had it the same way as your 2008 article in the first place.
Although “M+S versus M” is technically identical to “S+M versus M”, they are not subjectively identical terms because the former term generally requires more cognitive reparses than the latter — even the latter term “S+M versus M” generally causes a much higher cognitive load than does the term “A+B versus B”.
There is nothing inherently wrong with an “A+B versus B” trial provided that: the domains of A and B are the same; the units of A and B are the same; and the phase angle between vectors A and B is zero. In other words, it is a mathematically valid summation of the quantities A and B.
Furthermore, there is nothing inherently wrong with an “A+B versus B” trial in which A and B are totally orthogonal vector quantities (quantities that have a phase difference of ±90 degrees). However, when there is a difference in the phase angle between A and B it essential to correctly assign the placeholders A and B in each “A+B versus B” trial. E.g. echo cancelling equipment is an essential part of international telephony communications because, without it, it becomes increasingly difficult to speak fluently as the echo of our voice is increasingly delayed beyond 50 milliseconds.
If we conducted an “E+L versus L” trial on echo cancelling equipment in which
E=the canceller benefit
L=the transmission link audio quality benefit
then we have concocted a bogus trial in which the data can produce only the bogus — but statistically significant — positive result that E improves L. No, an echo canceller (E) does not, and cannot, improve the base audio quality of the link (L).
If we instead, correctly, conducted a reversed “L+E versus E” trial on echo cancelling equipment then we have concocted a very good trial in which the data will show very clearly how well each echo canceller (E) adapts to variations in transmission link audio quality (L). This correct usage of “A+B versus B” trial methodology (deployed decades ago) resulted in the awesome adaptive echo cancellers that are so good at what they do the vast majority of people are not aware of their presence in our global telephony networks.
Quackery relies on the bogus methodology of “Q+M versus M” trials (where Q=quackery and M=21st Century medicine), which cannot produce a negative result, and this is why quackery is an empire of (lucrative) ‘religions’ that are vehemently opposed to modern science and its methods.
Professor Ernst has demonstrated repeatedly that the correct trial methodology for the alt-med empire is not “Q+M versus M”, it is “M+Q versus Q”, and that only the results of the latter type of trial can produce meaningful risk:benefit and cost:benefit ratios. Why? Because alt-med (Q) is a synonym for the alternative-to-medicine empire that does not, cannot, and doesn’t care that it cannot, cure or prevent any known illness, in any species; whereas M is not perfect, but it is soundly based in empirical evidence and the scientific method. Quackery is stagnant anti-science dogma that has no mechanism for self correction, whereas medicine is continually self-correcting.