Massage for low back pain is an evidence-based therapy, despite the need for more study

Massage therapy is an effective treatment for low back pain: that’s the finding of a comprehensive review of the science, from a 2008 Cochrane review (Cochrane reviews are generally considered to be the most authoritative summaries of scientific research). The conclusions are strongly positive (emphases mine):

Massage is beneficial for patients with subacute and chronic non-specific low-back pain in terms of improving symptoms and function. Massage therapy is costly, but it may save money in health care provider visits, pain medications and costs of back care services. The effects of massage are improved if combined with exercise and education. The beneficial effects of massage in patients with chronic LBP are long lasting (at least one year after end of sessions).

Sounds pretty good, doesn’t it? But I’ve been burned by Cochrane reviews before — they aren’t all good. You can never take the abstract of any paper at face value: you simply have to read the fine print. There’s a great deal to consider about this paper, both good and bad. I’m going to examine it critically, but fans of massage should not feel alarmed: after a rocky start, this analysis eventually loops back around to a good-news conclusion.

Still, I have some concerns about these conclusions, and it’s important for me to be extra critical of science about massage therapy (see the sidebar). Despite the good news in this study, let’s look at the bad news first.

Comparing apples to orange cars: 13 completely different massage therapy studies

The glowing conclusions of the Cochrane review were based on just 13 trials (about 1600 participants) — quite a small number of scientific papers to review, and they were not generally of high quality. More than half were deemed to have “a high risk of bias,” and the studies were all over the map in terms of what they studied and how they studied it. It would be hard to imagine a more diverse group of experiments.

Reviews of science are looking for the common ground between studies. There’s not much point in combining data about different things: you want to compare apples to apples. This review was saddled with comparing apples to … orange cars. Almost everything was different in every study:

• the practitioner credentials and experience were different
• the massage techniques used were different
• the techniques used as a comparison to massage were different
• the types of low back pain being treated were different
• even how success was defined (outcome measures) were different

What a mess!

Massage techniques in these studies ranged from traditional Thai massage to “one 30-minute session of deep cross-friction massage” to “acupuncture massage.” One paper even used reflexology! That’s not reasonable: reflexology is not massage therapy, it’s massage-esque quackery. And — here’s a shocker — reflexology was the only kind of massage therapy that didn’t perform well.

Clearly, this was not a review with nice, apples-to-apples data. These 13 studies had almost nothing in common except that they were all experimenting with some kind of massage-like therapy for some kind of back pain. Generally speaking, if I encountered a mess of data like this in any other context, I would regard the results as nearly meaningless just because of this problem alone.

But please read on — this isn’t over yet.

More bad news: no truly good data

Although some of the studies were certainly stronger than others, not one of them produced truly good data. The best of data of the lot is merely “okay.”

When you want to know if a therapy works, you compare it directly to a fake treatment. That’s the bread and butter of evidence-based medicine: does treatment X work better than something known to be useless? And yet, out of all this data, only two studies hit the nail on the head and actually asked the question “Does massage work for low back pain?” by comparing massage to a sham treatment. (And they were not even the best studies, for other reasons. They were just the ones that had the right idea.)

Every other study either measured some outcome other than pain (pain quality, functional status, return to work), and/or compared massage to some other therapy instead of a proper sham (manipulation, mobilization, relaxation, exercise, etc), and none of them really answers the question “Does massage work?” For instance, if massage works better than mobilization, it’s nice to know that you might want to get massage instead of mobilization — but what if mobilization actually makes low back pain worse? Hypothetically? The apparent superiority of massage could be an illusion!

Comparing one poorly understood and controversial therapy to another poorly understand and controversial therapy does not really produce the most meaningful results.

But massage still wins: lots of positive results, and the most positive results in the strongest studies

Despite all of these problems, massage “wins.” Regardless of the study (excluding the reflexology one), massage performed brilliantly — no matter what kind of massage, no matter what kind of low back pain, no matter what it was compared to, no matter what outcomes were measured.

Best of all — the results were positive even in the higher quality studies.

So all the inconsistency is actually a blessing in disguise: if virtually any kind of massage can outperform basically any other therapy you compare it to, regardless of how you define success, that bodes well. It starts to seem safe to assume that well-trained and experienced therapists might do even better, maybe even a lot better.

But how positive was positive?

One of the biggest problems with the science of alternative medicine is that proponents of unproven therapies often try to make the most out of a small positive result, blowing it way out of proportion and acting like it’s a big deal (often for decades, long after plenty of subsequent evidence should have stopped the debate). When people want to know if a treatment works, they don’t mean that it works “kinda, sorta, maybe, a little bit.”

So it’s another good sign that one of the best studies of the 13, well-known to me for many years now, produced extremely good results. Massage therapy as delivered by well-trained Ontario therapists didn’t just work better than “exercise,” it was dramatically superior on each of three different outcome measures: patients had way more reduction in pain intensity and quality and much greater improvement in function, both immediately after treatment and still a month later. Booyah! That passes the impress-me test! (I am rarely impressed.)

This is precisely the opposite of relying only on the weakest science to prop up a useless treatment. Here we have weak studies showing positive results, and stronger studies showing even stronger positive results, and that’s what you expect from a treatment that “works” — when properly tested, it performs.

So, is massage therapy for low back pain “proven” to be effective?

Ha! Not by a long shot. There’s much more work to be done: dozens of studies instead of “a dozen.” Ideally all of them should be bigger — 100 subjects at least — with much better quality control and consistency in the type of massage tested. And no reflexology! But, unlike virtually all other popular low back pain therapies, the science so far is genuinely promising, instead of underwhelming.

Presumably this is why massage has, over the years, gotten a tentative thumbs up from many medical low back pain experts. For instance, Dr. Richard Deyo of Seattle was dismissive of all other manual therapies for low back pain in his excellent low back pain tutorial for the New England Journal of Medicine, yet he wrote that “promising preliminary results of clinical trials suggest that research on massage should be assigned a high priority.”

Several years later, the research still isn’t much better than preliminary, but the signs are even better than they were.