A large study of massage therapy for low back pain in the July issue of Annals of Internal Medicine (see Cherkin et al) suggests that massage may be helpful — “massage therapy improved function and decreased pain more than usual care.” This is consistent with the idea that back pain is often caused or complicated by “trigger points” or muscle knots.
The research has mostly been reported as a good news story for massage, but you may want to curb your enthusiasm because there’s also some genuinely bad news for massage hidden in there. In particular, the results also suggest that it doesn’t much matter what kind of massage — a result quite likely to ruffle some feathers, if anyone pays attention to that detail. I will pay attention, of course. I will report in detail on several neglected implications of this research.
Massage studies are generally rare, and the big ones can be counted on a hand. Hundreds of patients were tested by Seattle researchers Dr. Daniel Cherkin, Dr. Richard Deyo and several colleagues, and the size of the experiment alone makes it worth taking a close look at. It also had a number of other rare qualities and obvious improvements over other massage research:
And yet for all that, serious flaws still remained, and the authors acknowledge that it’s “difficult to determine the true magnitude of the benefits of massage observed in this trial.” A lack of blinding was the worst of them. It’s such a significant problem that it’s hard to know if the study can really tell us anything at all, despite its strengths.
Four hundred patients with “moderately severe” chronic low back pain, without a clear cause, were split up into three groups: for ten weeks, one group got weekly hour-long relaxation massages, another got more advanced therapeutic massage, and patients in a third group were essentially just paid $50 to do nothing in particular. Massage was provided by moderately-trained therapists2 with at least five years experience.
After ten weeks of massage — with a market value of about $500–1000 for each patient — about 60% of patients seemed to have about 30% improved function and symptoms from the starting point. However, there was no meaningful difference between the two massage groups — the difference was actually smaller than the range of uncertainty in the data. Patients who were left out entirely also improved — a very important point — but not quite as much.
A 30% improvement sounds good, but people in the “usual care” (no massage) group also improved, about 10%. On a pain scale of 10, massaged patients dropped about 2 points on average, a pain reduction of roughly 30%, while those without any massage dropped only about half a point. So the difference between massage and no massage was pretty small — just barely enough to be considered clinically useful. 30% is a noteworthy improvement from baseline, but far from a cure, not a whole lot better than doing nothing, and low bang for buck for real patients.
When the massaging ended, patients slowly but steadily lost their gains, while the never-massaged patients continued to slowly but steadily improve. By six months all scores were pretty much identical, with just a little advantage still remaining for massaged patients — they still had slightly better scores for function.
After a year, there were no differences left between any of the patients, and on average these chronic sufferers still had some back pain. Based on these results, the authors of the study came to the conclusion that
massage therapy may be effective for treatment of chronic back pain, with benefits lasting at least 6 months. No clinically meaningful difference between relaxation and structural massage was observed in terms of relieving disability or symptoms.
I think it’s a bit silly to describe a distinctly temporary 30% improvement in symptoms for 60% of patients as “effective,” and sillier still to say that the benefits lasted “at least” six months when the benefits were so reduced at that point that they barely register.
And then there’s the fact that improvements these patients seemed to have may well have been partly a mirage in the data. In fact, it’s almost certain …
This study lacked “blinding”: A key fact was not hidden from the participants, resulting in significant data pollution by their reaction. The patients who didn’t get any massage knew full well that they were participating in a massage study — they signed up for it and got paid — but also knew that they were not actually getting any massage therapy. (And free massage therapy at that.)
Ask yourself: How would you feel about that? Who wouldn’t feel a little disappointed and pessimistic?
Whether it’s effective therapy for back pain or not, people like massage. So these patients knew they had pulled the short straw, and probably expected to do less well — a perfect setup for slanted results. This is not a hidden flaw that I’m exposing: Cherkin and colleagues know it, and acknowledge that this problem could well be “making massage therapy seem more superior than it really is” in this study. This has been (charmingly) called a “frustrebo effect” — consisting of both a true negative placebo and “frustrated,” negative reporting — is a known problem with designing studies of popular treatments.3
Low back pain is notoriously sensitive to expectations.4 When you’re comparing a group of people who are consciously disappointed in their fate, you can pretty much count on their results seeming and/or actually being less good. Consequently, you could say that this study showed that not getting massaged was kind of a drag, and it can’t actually tell us if massage was actually better than just doing nothing.
Bear in mind that while massaged patients improved about 30% from the baseline, the non-massaged patients also improved — closing the meaningful gap between the two. The important question is not “How much did massage help?” but “How much more did it help than doing nothing?” That difference was very modest to begin with, and it’s a virtual certainty that the difference would have been even less if the un-massaged hadn’t known what they were missing. The frustrebo effect closes the gap.
Blinding is one of the pillars of a good clinical trial: if participants know too much, their hopes and fears and miscellaneous mental messes usually foul up the results. Generally speaking, people are quite alert and they can easily tell when they aren’t getting massaged for free. The only way to keep them happy about this is to make sure that they never knew it was a possibility in the first place. In this study, the patients needed to not know what they weren’t getting.
It’s really too bad that they didn’t not know.
An interesting feature of this experiment was that Cherkin et al compared the effects of garden variety relaxation massage — classic Swedish — with so-called “structural” massage. And what is “structural” massage, you might wonder? Good question!5
According the description, it’s a dog’s breakfast of several allegedly advanced massage techniques, all revolving around the dubious notion that low back pain is some kind of biomechanical failure or structural imbalance, correctable with just the right kind of pressures and manipulations of soft tissue:
Myofascial techniques are intended to engage and release identified restrictions in myofascial tissues. Neuromuscular techniques are used to resolve soft-tissue abnormalities by mobilizing restricted joints, lengthening constricted muscles and fascia, balancing agonist and antagonist muscles, and reducing hypertonicity.
Gobbledygook! A smorg of trendy and traditional massage jargon, most of it hopelessly vague, some of it mutually exclusive. The whole school of thought that massage should have a “structural” intention is still a major theme in the profession, but it’s debatable at best and debunked nonsense at worst.6 The idea of structuralism was taken to an extreme in this study, and perhaps deliberately, as we see in this final odd point of the description:
[Structural] therapists could recommend a home exercise consisting of psoas stretch to enhance and prolong any benefits of structural massage.
Including a stretch of this muscle as the only exercise to complement massage is an veeeery interesting choice. On the one hand, it’s ridiculous on the face of it. On the other, it’s also a savvy nod to the kind of clinical reasoning that is actually going on in massage therapy offices everywhere. For better or worse, psoas stretch is a common prescription, and a good representative sample of what supposedly “advanced” massage actually looks like out there in the real world.
And maybe that is a useful thing to test.
I would have been much happier if the experiment had also tested actually advanced massage techniques — as defined by me, since this is my fantasy — instead of a potpourri of vague and trendy ones. But, failing that, why not test techniques that are actually popular, however misguided they may be? Yes indeed, why not? And that is what we got: a test of just the sort of stuff that patients are likely to encounter in massage therapy offices in the wild.
And it didn’t work any better than the Swedish massage …
The results of the smackdown, according to the referees:
A course of relaxation massage, using techniques commonly taught in massage schools and widely used in practice, had effects similar to those of structural massage, a more specialized technique.
Here again Cherkin and his colleagues use language to describe their results that I find just slightly disingenuous and biased in favour of massage, and at odds with the data. “Similar” is not the right description. It was actually “no real difference at all.”
All that pretension. All those assumptions, psoas stretches, and lovely-sounding structural theories. All those expensive technique workshops those therapists went to, and all the extra money they charge real patients for their “expertise” to help pay off their investment in the workshops.9 It all added up to … nothing. They could have done relaxation massage instead and their patients would have been just as well off.
Maybe even better off. They would have spent less, for starters.
There were slight differences in results, but most of them actually favoured relaxation massage. The greatest of them was at the one-year mark: patients who’d gotten advanced massage actually scored a full point lower on the pain scale than those lucky Swedish massaged patients.
These are the kinds of data differences that should not be emphasized, however, because the wiggle room for error is actually larger than the measured difference — like trying to measure centimetres with a ruler that only has inches on it. The technically correct thing to say about the results is just that there was “no statistically significant difference” between the results.10
Still … it’s hard for emotional primate minds to ignore the fact that the data points were actually a little worse for advanced massage. And it certainly does help to drive home the fact that advanced massage was definitely not better. Maybe not actually worse. But clearly not actually better.
This study has been widely reported as a good news, “it works” story for massage. And the authors own conclusions sound pretty positive. Not so fast. As is ever thus, it’s complicated.
The results make typical so-called advanced massage really look bad, and they make the popular modality empires and structuralism as a paradigm look ridiculous. The technique gurus push and sell the idea that their methods are dramatically more effective than humble Swedish. If they were even half-right, these “advanced” therapists should have gotten results at least 50% better than their lesser-trained comrades — not just better by a statistically significant margin, but much better, impressively better, decisively better, undeniably better, argument-stopping better, better with bells on …
Instead, it’s like the New York Yankees accepted a challenge from a beer league softball team and couldn’t do better than a tie score.
The gap between the pretension and carefully measured results is a nasty condemnation of a huge chunk of an industry, at least half of all massage the way it is actually being practiced (probably much more).
This study has many weaknesses, and cannot actually tell us if “massage works” — the other bad news — but if nothing else it has certainly produced extremely strong evidence that the major advanced massage modalities do not actually work at all, and are just not worth the money. It’s a certification racket, and massage therapists need to get just as cynical about it as they probably already are about Big Pharma.
On the bright side? Relaxation massage is relatively good stuff: cheaper, more accessible, and there’s nothing “just” about it. A good Swedish massage is high art, and not nearly as simplistic as it has been portrayed by therapists who figure they’re too good for Swedish. In particular, relaxation massage places a much higher priority on addressing the human nervous system. I think that guiding principle may well prove in time to be a more advanced method than yarding on people’s tissues with the barbaric intention11 of actually physically changing them.
Not that relaxation massage actually “won” the contest — that would have been interesting — as the benefits of both styles were roughly equal and therefore equally unimpressive.
I cannot actually agree with the authors that their massage recipients got “clinically meaningful improvement” (especially at six months). It is possible that they had clinically meaningful improvement, but it is by no means certain. The problems with the study make it impossible difficult to conclude that any kind of massage actually worked. Indeed, when you get roughly equivalent results from quite different treatments, it tends to suggest that the results weren’t due to anything unique or specific that was being done.
It is also possible that improvements in pain and function are due to nonspecific effects, such as time spent in a relaxing environment, being touched, receiving care from a caring therapist, being given self-care advice, or increased body awareness.
“Nonspecific effects” is an important concept here. Nonspecific effects are the many potential effects of being treated and cared for that occur with any treatment, as opposed to effects that only occur when a specific treatment occurs. In particular, non-specific effects tend to be related to the interaction between a patient and a health care provider, because any kind of care — nothing specific — has some therapeutic effects.
In short, people get better (or claim to) when they get some compassionate attention.
A massage appointment is nonspecific effect nirvana. The entire point of a good massage is to provide a great interaction between patient and therapist. A massage patient is at the luxurious centre of attention, being cared for in a way that is arguably the single nicest (nonsexual) experience any human can have. Nonspecific effects are dialed up to 11 in this environment. (Unpleasantly painful massage is a complicated exception.)
That these effects exist and are generally optimized in this situation is hard to deny. It’s more a question of how much of the benefits of massage can be attributed to them. Most? All?
The data produced by this experiment can’t tell us, but if it’s “most or all” then these are just the sorts of results you’d get: a wide gap between satisfied massage patients and disgruntled un-massaged patients, and no difference between relaxation massage and alleged expert techniques.
Trigger points were not named in the description of “structural” massage: an interesting exception. Many of the therapists probably considered what they were doing to be (or include) “trigger point therapy.” And yet it was not singled out as a concept, which is surprising given that trigger points is probably the most popular idea in massage therapy. It was the only popular massage concept that wasn’t named. I wonder why?
This study cannot tell me much about trigger point therapy, but I do have a pretty good idea why relaxation massage was no loser: good trigger point therapy is actually relaxation massage.
I have taught readers for many years now that trigger point therapy must be an extremely humble, experimental, and relatively gentle process of artfully messing about with a variety of pleasant sensory inputs. Few human experiences are more satisfying than skilful stimulation of those mysterious perfect spots that we crave to have pressed. Time and again over the years, my “knots” got a more pleasing workout from a simple Swedish massage than from the no-pain-no-gain proddings of trigger point therapy as it is usually practiced.
Unfortunately, most massage therapists still think in terms of trying to physically change patients — a kind of wrestling match with their tissues, where just the right moves will banish a pathology from the meat. Unfortunately, trigger points are poorly understood and almost certainly not what they seem to be, so there really can be no formula for treatment, and the simplistic attempt to beat up trigger points (and patients) is a bad idea in several ways. It’s wiser — and more science based — to simply provide a rich and pleasing sensory experience that may, perhaps, result in the organism changing the tune it is playing.
Failing that — and it certainly does often fail — the patient at least has a good experience.
Drs. Cherkin and Deyo and their colleagues seem to have a case of acute rose-coloured vision. Much like their 1999 study of acupuncture,12 their conclusions are a fair bit more glowing and optimistic than the data seems to support. Describing massage as “effective” for “at least 6 months” sounds like they are talking about the results of some other study with much better and more certain results! At best, even if we could trust this data completely, it showed only modest and temporary benefits to quite a lot of expensive massage therapy. But we truly can’t trust this data: those apparent benefits may been mostly or entirely due to the acknowledge, obvious and unfortunate “frustrebo” of the patients who got nothing. We have to assume that the benefits of massage were not actually as strong as they appeared to be here, which almost certainly reduces it to a clinically trivial level.
I am a science writer, former massage therapist, and I was the assistant editor at ScienceBasedMedicine.org for several years. I have had my share of injuries and pain challenges as a runner and ultimate player. My wife and I live in downtown Vancouver, Canada. See my full bio and qualifications, or my blog, Writerly. You might run into me on Facebook or Twitter.
If you found this article useful, you may also be interested in some other articles I’ve published:
More than 600 participants were either given standard acupuncture treatments or simulated acupuncture. Although this study has been widely reported as if it was a controlled comparison of acupuncture to “standard medical treatment” for back pain, in fact it is not controlled (or blinded), and does not have the power to prove that acupuncture works for back pain.
The apparent difference between real and fake acupuncture they observed was minor. Nevertheless, the authors are excessively friendly to acupuncture and declare it to be “effective” in their conclusion in spite the obvious poverty of the data. In particular, they gloss over the damning implications of their most important finding: what little effect they think they found had nothing at all to do with needle placement. Acupuncture means nothing if needle placement doesn’t matter. The interpretation of Dr. Steven Novella is much more sensible: “The only reasonable scientific conclusion to draw from this is that acupuncture does not work.” For Dr. Novella’s meticulous and expert analysis, see Acupuncture Does Not Work for Back Pain (Part I).BACK TO TEXT