Exercise for chronic back pain: The beige trouser effect?

Most commonly used exercise therapies for back pain are aimed at having an effect on some mechanical or tissue based aspect of spinal function. From range of motion exercises to muscle balance, endurance or strengthening exercises the (not unreasonable) rationale is that back pain is associated with abnormal spinal function – address that with exercise and back pain should improve.

The problem is that, whichever way you slice the data, the effects of exercise are small to moderate at best (kind of the therapeutic equivalent of beige slacks, better than nothing but generally underwhelming).  It doesn’t seem to matter what type of exercise you do, comparative trials of different types rarely demonstrate a difference and systematic reviews have not clearly demonstrated a type of exercise that is superior. Many argue that the problem is with the data not the treatments; the lack of targeting treatments to the group of patients for whom it is appropriate means that the true effect is washed out by all those people in the trial for whom the treatment was inappropriate or ineffective. But you all know this already and I am repeating myself.

A new review out of Switzerland, just published in the European Spine Journal has taken the issues of exercise therapy for CLBP and subgroups and looked at the data in a different way. If there are specific subgroups of patients for whom exercise regimes work the way that they were designed to, then any improvements seen in trials should be correlated with a change in an appropriate measure of spinal performance (for instance improved range of motion for ROM exercises, or improved spinal endurance or strength for those kind of regimens). So they sourced all rehab trials that incorporated exercise and a measure of spinal function and reported the results of correlation analyses between clinical outcomes (pain and disability) and physical function.

Their results are fascinating. The overwhelming majority of those trials that looked found no correlation between clinical outcome and spinal functional performance, and the few that did found weak or inconsistent correlations. So when people with CLBP feel better after exercise therapy, that change does not seem to be reliably reflected in improvements their spinal function.

It could be agued that the researchers in these trials just measured the wrong parameter of spinal function. I wonder what the correct measure would be? Also these studies might not have adequate power to detect a correlation, particularly if the subgroup of responders is small, but that in itself would be a fairly powerful indication of the limitations of the therapy.

How should this be interpreted? Directly it infers that where exercises help it’s not by the traditional mechanisms by which we thought they might. It lumps the effects of exercises, at least for now, under the label “non-specific”. Indirectly it suggests that if there are specific subgroups for whom exercise therapies have benefits, then the improvement in those subgroups was not likely due to the suggested “active ingredient” of the exercises given. I think it pulls the rug out a bit further from under the feet of the subgroup argument because in terms of specific effects there seem to be no funky britches hiding in the sea of boring beige.

About Neil

Neil OConnellAs well as writing for Body in Mind, Neil O’Connell is a researcher in the Centre for Research in Rehabilitation, Brunel University, West London, UK. He divides his time between research and training new physiotherapists and previously worked extensively as a musculoskeletal physiotherapist. He also tweets! @NeilOConnell

He is currently fighting his way through a PhD investigating chronic low back pain and cortically directed treatment approaches. He is particularly interested in low back pain, pain generally and the rigorous testing of treatments. Link to Neil’s published research here. Downloadable PDFs here.



Steiger F, Wirth B, de Bruin ED, & Mannion AF (2011). Is a positive clinical outcome after exercise therapy for chronic non-specific low back pain contingent upon a corresponding improvement in the targeted aspect(s) of performance? A systematic review. Eur Spine J PMID: 22072093  [Epub ahead of print]





  1. Lucky for me I don’t see, feel or hear funny tones in a correspondence, maybe because Ozzie language is a foreign language to me. Using a lot of Mr. Peter O’Sullivan stuff wrapped around a nice NOI paper, using dear Mr. Butlers sliding nerve techniques with empathy I seem to be able to help a specific group of non-specific chronic low back pain.
    Not very scientific though. The outcome is seldom measured in a better r.o.m. of the spine. Not interesting. Seeiing good movement patterns in a graded excercise learning program is the best outcome for me and I have found no objective measurement(s) for that. Than questionaires seem the best alternative, but those seem to be just as subjective. Sometimes we are just buying time for the client, but I truly hope it is (much) more than just that. These findings of Phd’s though do me no good in that department. So go and do more research!

  2. Neil O'Connell says:

    Dear Ken and Paul,

    I reckon it is worth keeping the tone of the exhanges here as polite and respectful as possible. Your clinical findings are interesting Ken but when there is a broader body of higher level evidence suggesting the back extensor strengthening does not offer large effects (e.g.http://www.ncbi.nlm.nih.gov/pubmed/18298356) we would have to be very cautious in accepting that observational clinical experience as robust. The individually tailored treatment issue with RCTs is to my mind something of a red herring. Depending on the research question being asked, different trials in back pain tend to involve a variable level of discretion to the treating clinician to tailor their treatment to the patient and more recent trials tend to be more pragmatic in that sense. Still we haven’t seen great effect sizes.

    Back pain, be it acute or chronic will usually appear to improve in the clinic and in trials(see here http://bodyinmind.org/research-into-back-pain-treatments/). It is reasonable to expect that if exercise therapies do what they claim to on the packet then we will see improvements of reasonable size that are correlated with improvments in a fair measure of function. This particular review provides evidence against this prediction.



  3. ken cameron says:

    Your condescending reply does you little credit.
    What I suspect is going on is that evidence based researchers such as yourself get all wound up in their own little world and forget the big picture.
    Remember that the application of science to healing people is an art.
    Systems need to be tailored to individuals in treating persistent pain and your colleagues at noi have excellent information to this effect.
    Pragmatism , experience, and application of current knowledge using modern pain science including active physical therapy has a significant effect on management of all types of persistent pain whether you like it or not
    Take off your beige trousers and try perhaps a kilt, it might just work.

    Paul Ingraham Reply:

    Ken, I apologize if my tone seemed condescending to you: truly, that was not my intent. I was earnestly answering your question, proposing a reasonable explanation for what you’ve observed. Sure, there was some critical toughness to my reply, but I assumed that you submitted your observations for actual comment and critical analysis…not flattery.

  4. ken cameron says:

    Interesting article ,bizzare conclusions
    I am a gp a full time 30 year plus clinician with a background in mental health and a long term interest in persistent pain.
    In the last 12 months I have been sending my clbp patients for physical therapy via the medex system which is based on back extension against resistance.
    The results have been quite startling with 80 percent of the 110 patients referred reporting significant improvement in terms of pain reduction and function at 6 weeks ,sustained at 6 months. I can correlate this to reduced medication use and clinical and computerised measurements.
    Any comments?

    Paul Ingraham Reply:

    Ken: People tend to recover from back pain? I don’t suppose you’ve compared those results to sham and no-intervention therapies? Anything less is really worth talking about, really. In fact, it’s worth not talking about. We are drowning in low quality research noise and anecdotal evidence already.

    Paul Ingraham Reply:

    Oops: “anything less is really NOT worth talking about.”

  5. The study is looking at chronic low back pain which we know is a complicated beast. I agree with Anne’s comments above. When dealing with chronic low back pain exercise is part of the solution but it needs the psychological support backing it up to over come limiting beliefs.
    Has there been research comparing exercise for clbp with a group that has exercise and some sort of psych support?

  6. Research results which you mention for exercise in CLBP seem to show that focusing on pathology and trying to find scientific proof of the effect of exercise on CLBP is as difficult as trying to relate symptoms directly to Xray and scan reports. There seems to be no direct connection.
    Pain is so complex, and persistent back pain so tied up with attitudes to pain, social circumstances, fear of movement, normal attitude to physical activity and hurt, that it is so difficult to separate out the psychological benefits of doing exercise from the physiological. it seems to me that exercises for CLBP often are just a tool to overcome activity fear and raise activity levels. Being physically active is the answer to reducing the back pain perhaps rather than any specific exercise. The problem is that people look for and must be given explanations for treatment suggestions (informed consent), so we all try to present logical explanations for the exercises we generally have found to be helpful in a purely anecdotal way. I teach exercises highly recommended for particular presentations so that I feel I am offering the best quality treatment I can. Sometimes it is just a matter of trying exercises which seem logical and seeing how the patient responds and altering them according to that.
    I also think that improvement depends on the whole package including a cognitive behaviour therapy type approach in talking with the patient and building their confidence in managing their pain.
    It seems very difficult to get a true outcome when you separate aspects of the whole therapist patient connection.
    (I am not a researcher, but just a regular hands on physio who finds your discussions interesting)

    Neil O'Connell Reply:

    Hi Anne,

    I think your points are all reasonable. However even multimodal approaches don’t do so well and as we’ve already covdred here behavioural therapies do very poorly in trials despite these therapise being individually tailored by design.(http://bodyinmind.org/chronic-back-pain-behavioural-treatments/).

    As a clinician (which I also am) it is hard to know what to do with the current state of the evidence, indeed there is an emerging case for doing nothing beyond triage and advice (http://bodyinmind.org/research-into-back-pain-treatments/). This would represent a revolutionary downsizing for a burgeoning industry of course but the case against it is currently pretty weak.

    The epidemiologist Heine Raspe has an excellent quote to this effect (I am doing this from memory as I am out of the office so apologies for errors):
    “It is an promiosing hypothesis that the demedicalization of low back pain may do more to alleviate suffering than all of the combined investigations and treatments available for this condition.”

    I will get the real quote tomorrow and post it as it is more eloquent than my memory. This is a radical change of mindset but not outrageous given the base.

    Neil O'Connell Reply:

    And as promised the actual quote/ reference:

    Heiner Raspe (2002) How Epidemiology contributes to the management of spinal disorders. Best Practice & Research Clinical Rheumatology 16: 1: 9-21

    “It is a promising hypothesis that de-medicalization of non-specific back pain may eventually lead to less overall suffering, chronification and social disability than all of the medical, both diagnostic and therapeutic, interventions that are currently employed”.

    Pretty bold stuff, beautifully put.

  7. Jeffre, you wrote, “most pts do not care if their back extensor endurance is up to par.” I beg to differ!

    In my experience, patients are regularly and persuasively told by professionals that various measures of spinal performance and function are critical factors in recovery from low back pain, acute and chronic — and that they are in pain precisely because they are not up to par. Or that they are recovered because they are. I wish I had a buck for every time I’ve heard it from a patient, and another buck for every time it was closely tied to advice received from another professional.

    It seems to me that proponents of exercise for back pain often claim spinal function as an etiologic and therapeutic mechanism, and patients have been buying it for decades now. In fact, I can hardly think of a more widespread belief about low back pain.

  8. Thanks for the review of the article. After a quick glance at the article here are some thoughts that I have: 1. Isn’t the whole idea of stabilization exercise is the motor control part? 2. I thought that it was common knowledge that there is no link between strength and LBP? 3. It does not appear that any of these studies classified the pts using the Treatment Based Classification system that has been used in a good bit of PT literature. 4. I wonder why the studies did not use the ODQ/GROC/PSFS for outcome measures which are much more common in the clinical setting. 5. As a clinician out in the field, pts feeling better is not bad thing and most pts do not care if their back extensor endurance is up to par. 6. Finally, I think you are right that these studies are looking at the outcomes. Pain, functional outcome measures and pain/pt education would be better measures.


    Neil O'Connell Reply:

    Hi Jeffre,

    Thanks for your comment. The argument that the original studies looked at the wrong measure of spinal function has some legs except that one might expect that if motor control improves back pain then that might be reflected in improvements in ROM (although perhaps not strength). The studies all used clinical measuires like the RMDQ or the ODI. It was those measures the review sought to perform correlation analyses with the measures of spinal function. With the treatment based classification thing – no particular form of subgrouping could currently claim to strongly backed up by consistent quality published data. Finally – patients feeling better is no bad thing regardless of spinal function. The problem is our best data suggests patients with CLBP on the whole don’t tend to feel much better after exercise therapy. That is the big concern.

  9. So, in other words, even when patients felt better, but they didn’t work better? In terms of of spinal functions supposedly related to the type of back pain they had? Intriguing.

    At least three times in other places over the last day, I’ve seen other people call this article and this research “underwhelming” … but simple consistency with the null hypothesis never is very whelming. And that’s the point. The null hypothesis is holding its ground here: we’re not overwhelmed by any evidence for the claim. Where’s the support for an effect that, according to the claim, should be robust? Or at least detectable?!

  10. Neil,
    Good points. In my humble view, exercise is part of the picture. Exercise is a behaviour, chosen on the basis of a belief system, self-efficacy, expectation and really whether it makes sense to the patient. Changes in movement, sense of the body, fears of ‘damage’ and other expectations of pain will influence the way the individual moves. Setting the patient up in the right mindset, with good understanding of their pain and what they are going to do physically to help normalise movement and sense of self, I think is helpful. A good experience usually develops confidence and self-belief. An ‘it’s ok’ moment.
    I break down the exercise or physical activity aspects of the programme to the specific movements that are addressing specific identifies issues (limited movement, guarding etc), general exercises and those that are a good idea for the rest of the time, that huge chunk of the day often forgotten. Until it hurts from sitting too long. And the. It’s too late in many cases. I encourage proactive rather than reactive.
    Just some thoughts.

    Ps/ I like the beige analogy. It makes me think of the allegro and words like ‘nice’, ‘lovely’ – they’re ok, we know what they mean but as you say, underwhelming.

  11. Thanks Neil- as always an insightful and critical post.

    Out of interest which subgrouping strategies are you referring to? My perception of most up to date thinking on subgroupings was that they were designed to help us analyse movement and use movement behaviour as a way of changing beliefs.

    Obviously I realise there are many other ways to change beliefs!!!!



    Neil O'Connell Reply:

    Hi James,

    I think the argument applies to most subgrouping strategies. If there is a subgroup of folk in there who respond much more strongly, be they defined by structural diagnosis or movement pattern, one would expect to find a correlation between clinical outcome and spinal function. The absence of this correlation suggests that clinical improvement doesn’t have m,uch to do with spinal function and isn’t reflected in altered spinal function (assuming that the trials measured the correct paramter of spinal function)

    James T Reply:

    Cheers Neil

    How do we than start to interpret Kjartan Vibe Fersums’ Phd findings? Is the argument that it wasn’t the sub-classification that was significant but the cognitive element of management? Would that mean that we should compare these findings to Cognitive input with no exercise?

    I would stress I am not trolling!! Just a clinician who is interested!


    Neil O'Connell Reply:

    Hi James,
    I’ll have to have a good read and won’t get a chance until Monday, but if you are referring to the 2010 systematic review the abstract does not suggest a big effect or a lasting one. Will look in more detail.

    Neil O'Connell Reply:

    Hi James – weekend over! The Fersum review found very little data that asked the subgroup question and that which they found was a mixed bag of quality that subgrouped in a variety of different ways. There needs to be more data, and data which subgroup in similar ways for a pooled analysis to be interpretable. The results don’t really give us any idea of the best way one might subgroup people so I don’t feel thay can guide practice.

    I hope that helps.

    ed Reply:

    I have plenty of history with CLBP. Years of it. MRIs, PT, stretches of short term disability, lots of oxycontin, you name it, I’ve been through it. For nearly 4 years now (I’m 47) I’m as closed to being cured as I can be. The trick – running, and lots of it. No medical suggestion to do such a thing, in fact quite the opposite along the way, but it has been to me a genuine miracle, a life changer. Not to say I don’t still have periodic episodes, but they are never as severe as they have been in the past, but the frequency is greatly diminished and the duration of an episode is much shorter. Running while my back is killing me seems to actually help get things straightened out sooner, whatever that means.