Yoga vs. stretching for lower back pain


As of September 2017, new Sweat Science columns are being published at Check out my bestselling new book on the science of endurance, ENDURE: Mind, Body, and the Curiously Elastic Limits of Human Performance, published in February 2018 with a foreword by Malcolm Gladwell.

- Alex Hutchinson (@sweatscience)


I tend to post a lot about studies that find no benefits from traditional static stretching. Does that mean stretching has no benefits? No — it just means that the benefits are hard to quantify. So to be fair and balanced, I figured I should mention this recent study from the Archives of Internal Medicine, which suggests that stretching may be helpful for lower back pain (press releases here and here).

The study was actually designed to test whether yoga helps back pain. They compared a 12-week yoga program to 12 weeks of stretching (chosen to have a similar level of physical exertion), or 12 weeks reading a self-care book. Both yoga and stretching were better than reading the book at improving pain and function; there were no differences between yoga and stretching.

Now, I can’t help pointing out that the study isn’t immune to placebo effects. The assessments of pain and function were done with telephone interviews, and relied on subjective reports from the patients. And let’s be honest: the suckers who were randomized into the “self-care book” group knew darn well that they got the short end of the stick! So I don’t view this as strong evidence of a mechanistic relationship between stretching and back pain (i.e. that the back pain is caused by tightness in some specific muscle, and stretching releases the pressure to eliminate the pain). But that’s kind of beside the point. The stretching made people feel better — and for a very simple, low-cost, low-risk, uninvasive intervention (unlike, say, surgery), that’s a good enough outcome.

11 Replies to “Yoga vs. stretching for lower back pain”

  1. That sneaky “frustrebo” effect might well account for the entire apparent benefit here. This is also the second Cherkin trial lately afflicted by it: it was a big deal in a recent big study of massage for low back pain.

    Also note that the effect size here was modest, even if it was real: only just detectable, really. A 2.5-point difference on subjective scale to 11 is a pretty weak sauce. Some people might even say that’s too much hassle to go to a bunch of yoga classes for that. Then there’s the fact that yoga isn’t actually risk-free, or cost-free.

  2. @Paul: Thanks for the link — I enjoyed your detailed and much less diplomatic takedown of the massage study! 🙂 (And I like that term “frustrebo,” which I hadn’t encountered before.)

    And of course, I agree with your assessment of the size of the effects: “Some people might even say that’s too much hassle to go to a bunch of yoga classes for that.” I’d certainly agree with that.

    In this case, the researchers hoped that their study would show that yoga helps. (“We expected back pain to ease more with yoga than with stretching, so our findings surprised us,” one of the researchers says in one of the press releases.) Stretching was only supposed to be the control group — the one that avoids the frustrebo effect. The fact that stretching and yoga turned out to produce the same results should really be viewed as a null result.

    But still — and this is a question that I wrestle with for a lot of the questionable performance- and health-enhancers that I debunk… If an intervention produces a small benefit at minimal cost (which is true for stretching, if not yoga), does it matter if it’s basically just a product of hope, belief and self-deception?

  3. I agree with you guys, the yoga for me personally seems like too much trouble. Yet, at the same time, I’ve never had chronic low back pain and I hate doing yoga.

    One thing that should be kept in mind when interpreting these studies is that there are a number of causes for low back pain (hypomobility, hypermobility, CNS sensitization ect..). If your low back pain is originating from an instability (or at least a hypermobility), then why would stretching help? By the same logic, it would seem that the hypomobile painful back may be more likely to respond to the stretching regime. Unfortunately, these two types of LBP average to the typical “almost no improvement” result.

    It comes down to classifying low back pain into different subsets before accurate conclusions about specific interventions can be drawn. This is where the research is headed…but we are still trying to figure the best classification system out:

  4. Indeed, I’m not all that diplomatic. 🙂

    Yes, I would also expect the stretching group to be pretty much a frustrebo free zone. However, the reading group was likely afflicted with it at least a bit, those poor souls consigned to a rather dreary arm of the study, deprived of either a lovely stretching or yoga experience. They might well have reported more negatively and/or actually had a slightly worse outcomes … thus creating the illusion of small benefit to both stretching and yoga in comparison. If frustrebo accounts for even a portion of the reported effect, it pulls it below clinical significance for sure.

    Certainly the lack of difference between yoga and stretching is the real headline here, but I think it’s almost as noteworthy that they are both equally damned with faint and illusory praise.

    But supposing yoga and stretching really did help, a little: a small positive effect, based on a stew of nonspecific effects … does it matter that it isn’t quite what it seems? Not a great deal — the stakes are relatively low here — but I think it does matter a little. I think we shouldn’t underestimate the tyranny of “shoulds” (i.e. “I really should do more stretching”) and the slow-motion havoc caused by chronically barking up the wrong tree for solutions to a nasty chronic pain problem. A little might be gained by participating in a yoga class and assuming it’s good medicine, but then again it’s not much to lose either … and it may well be repaid in full in other ways. For instance, nonspecific effects can also be based on education — rational, informed confidence can be quite liberating and “therapeutic” in itself.

  5. Thanks for the comments, guys.

    @Sean: Agreed. It’ll be no surprise to you that Stuart McGill said essentially the same thing when I interviewed him for an article I wrote last year on strength training for back pain. As he put it, “There’s no such thing as ‘non-specific’ back pain.” It always has a cause — “non-specific” just means the cause hasn’t been diagnosed yet!

    @Paul: These are all great points. Certainly, my personal answer in my own life to that rhetorical question (does it matter is the effect is based on self-deception?) is yes, it matters. And you make a strong case that it should matter to everyone. I guess I’m just a little hesitant to overstate the firmness of my own (or the scientific community’s) knowledge in these cases. I definitely don’t think there’s much of a case in favour of stretching (or yoga) to help back pain, but neither do I think there’s a water-tight case that it CAN’T help. So I’m a little more willing to say “Well, maybe — if you think it helps, go ahead” than I am for, say, Power Balance bracelets, where there’s no such ambiguity.

  6. Chuckle: I almost used Power Balance as a comparison myself! 😉

    Agreed, there’s no slam dunk dismissal of stretching/yoga for back pain here, with numerous other reasons why someone might be happy/willing to mess around with it, not least of which is that it can simply be really pleasant. Mostly I just want low back pain patients to know that it’s no big deal — they don’t have to feel like they’re really dropping the ball if they’re not in a yoga class three times a week. It’s shocking how often I’ve come across that. Low back pain sucks. But low back pain plus unnecessary guilt? Worse!

  7. @Paul Ingraham

    I once heard about a professor in a university hospital who said that the secret of a good clinical trial is knowing when to quit.

    I assume you have noticed, that the 2005 study stopped at 1 year, and found that after 1 year, no significant difference remained. The new study stops at 26 weeks. On other words: stop while the effect is still significant.

  8. @Paul Ingraham
    Uhh… perhaps you haven’t noticed, since it’s the other way around. The 2005 study stops at 26 weeks and the 2011 study stops at 1 year.

    Point about cutting of while significant stands. Evil suggestion of deliberate cutting off not.

Comments are closed.