Should we turn away people with CRPS?

Two foods I love eating regularly (and that’s probably not a good thing for the latter) are tuna and ice cream. Thankfully, I have a firm grasp of the concept that some things just weren’t made for mixing.

You might think this is a funny way to start a blog post on CRPS… but I promise you the Adelaide heat hasn’t fried my brain. I contend that my tuna-ice cream metaphor might help us to understand why some interventions treating CRPS may not be working.

We recently published a systematic review and meta-analysis that looked at the effects of graded motor imagery (GMI) on chronic pain, where the majority of participants in the included studies had CRPS [1]. By pooling the results of two RCTs [2][3] we found positive results for GMI in reducing pain. Our results, however, are in stark contrast to those found in a recent clinical audit of CRPS patients using GMI, done by Johnson et al. [4]. In this study, participants reported very little improvement in pain scores and only minimal improvements in function following the intervention. (See our articles page to download PDFs published by the BiM team.)

That this study did not replicate the findings of the previous RCTs may be a result of differences in methodological rigour (for example, see what Simon has written here). Participants in this study seemed to not complete GMI, not do GMI properly, or do GMI as part of a variable programme. The fact that participants ‘received GMI, in conjunction with a range of other “best practice” physical and psychological interventions’ brings me back to my tuna-ice cream metaphor. Just as tuna and ice cream are nice foods when eaten separately, we have reasonable evidence to support the use of GMI and other ‘best practice’ interventions [5] to treat CRPS, when used as separate treatments Together though, I’m not so convinced.

Sometimes using a combination of treatments can be counterintuitive to one another. For a patient with CRPS that experiences catastrophic pain with light touch, tactile discrimination training (TDT) may not be an appropriate intervention. Because GMI is an intervention that doesn’t require tactile stimulation, it provides a step back from something like TDT. While there is good evidence to suggest these two interventions work when used separately, the combination of the two doesn’t equal double the benefits.

Of the thirty-five participants in the Johnson et al. paper, only eighteen completed the three stage GMI protocol; the other seventeen participants completed the first one or two components of GMI, either as stand-alone treatments or in conjunction with other interventions. All thirty-five participants were statistically analysed as a group.

After reading through the study a few times, I came to one bold conclusion – if GMI associated with a range of other interventions doesn’t work, then perhaps all of the interventions being done at these centres don’t work? If this is the case, then perhaps we should try doing nothing? What if we turned away patients with CRPS? See what happens to their NRS pain data when they go untreated? Of course, my common sense (and moral grounding!) tells me this wouldn’t end well either.

All in all, I think the conclusion of this study may appear a little precocious. To me, the results say that tuna-ice cream is bad, tuna milkshakes are bad, tuna chocolate truffles are bad, and rum and tuna are all bad. However, they’ve still come to the conclusion that tuna itself is bad. If we are going to accept the possibility that tuna is bad, then we also need to accept the possibility that ice cream, milkshakes, truffles, and rum might also be bad.

Jane Bowering

Jane Bowering Body In MindJane has an addiction to Vegemite, loves ice-cream, and tuna (but only dolphin-free!), and has just started her PhD at BiM. After finishing her Physiotherapy degree with Honours and working clinically for a year, Jane is now back into pain. Researching, not inflicting, that is. She is interested in pelvic pain, but the rest of her PhD story is yet to unfold.


See our articles page to download PDFs published by the team.

[1] Bowering KJ, O’Connell NE, Tabor A, Catley MJ, Leake HB, Moseley GL, & Stanton TR (2013). The effects of graded motor imagery and its components on chronic pain: a systematic review and meta-analysis. The journal of Pain, 14 (1), 3-13 PMID: 23158879

[2] Moseley GL (2004). Graded motor imagery is effective for long-standing complex regional pain syndrome: a randomised controlled trial. Pain, 108 (1-2), 192-8 PMID: 15109523

[3] Moseley, G. (2006). Graded motor imagery for pathologic pain: A randomized controlled trial Neurology, 67 (12), 2129-2134 DOI: 10.1212/01.wnl.0000249112.56935.32

[4] Johnson S, Hall J, Barnett S, Draper M, Derbyshire G, Haynes L, Rooney C, Cameron H, Moseley GL, de C Williams AC, McCabe C, & Goebel A (2012). Using graded motor imagery for complex regional pain syndrome in clinical practice: failure to improve pain. European journal of pain, 16 (4), 550-61 PMID: 22337591

[5] Daly AE, & Bialocerkowski AE (2009). Does evidence support physiotherapy management of adult Complex Regional Pain Syndrome Type One? A systematic review. European Journal of Pain, 13 (4), 339-53 PMID: 18619873


  1. stuart miller says:

    Jane, what is your experience with tactile discrimination training ? In terms of the framework of how we perceive the world ( Moseley’s on reconceptualization of pain and others) the idea of introducing a threatening input to an over-protective system (or region) has to be done delicately. Since we are so visually driven (and this may have developed due to a need for quick recognition of threats (ie predators – big cats, snakes) – dampening the input to this system with GMI is wonderful as long as the patient and therapist (even with outside influences) have a clear understanding of the framework of graded progression of threat to the CNS- making the objects for TDT as ‘non-threatening’ as possible (favourite objects or simply small soft and larger soft object; explaining it in terms of higher level processing of position sense) makes sense. As well, in terms of our other senses (hearing, smell) making the overall environment conducive to again re-exploring the world (tentatively at first – OK+, OK-) makes sense. Is the language of explanation critical ? (I am always looking at ways to make it non-threatening especially for a lot of the patients that I see in which English is their second or third language – I use the pictures from Explain Pain book a lot but it is the meaning for the patient in explaining a strategy that they can use that I think is most important ). So, in summary, I think it is an understanding of food (in view of your tuna metaphor) and the context that is critical. Dampening threatening visual input first seems to make the most sense but I look forwards to yours and others insight.

  2. stuart miller says:

    Thanks for this. What I was alluding to was that the 2 centres in Britain that were evaluated in the Johnson Study seem to be the main centres (with World Class knowledge or centres of excellence) for treating CRPS (as well as a lot of other pain conditions) and due to this, many patients had to travel relatively long distances to get there and were receiving follow-ups every 1 – 2 weeks (likely due to long wait-lists and busy caseloads ). What would make sense is to train ‘satellite’ centre staff ( PTs, OTs, ? Psychologists and Nurses) throughout Britain (with input from the staff with expertise in treating CRPS) so that patients would have convenient access for frequent and consistent follow-up. I realize this is a bit of an undertaking but it would be patient centred. The UK guidelines for treatment of CRPS are more than a little unwieldy (unlike the Netherlands) so a lot of work in terms of consistency of approach would need to be done (I realize that the treatment would have to be tailored to the individual patient but a consistent framework is helpful). The chronic pain centres would still treat patients with chronic pain but patients with CRPS may be seen sooner and more frequently (especially in early stages) before things ‘set in’. Just a thought…open to feedback.

  3. stuart miller says:

    Jane, what are your thoughts on ‘Centres of Excellence’ (for patients in which the stress response is important) that are ‘inaccessible’ or extremely challenging for patients to access. It would seem that a concept of training the trainers in more accessible areas with input from ‘expert’ clinicians so that more frequent follow-up (possibly daily without the need for inpatient admission) would be desirable. With the Research studies, the fact that they pay patients may factor into the success but I think it is the overall convenience and consistency of confident treatment that lends itself to success – this also lends itself to the therapeutic alliance. Please advise.

    Jane Reply:

    Hi Stuart,
    I hadn’t heard of the Centres of Excellence idea, but have just looked it up – I think I wasn’t aware of it because it appears to be a US initiative only. The idea I like though! I have a couple of points I’d make of the Centres though…

    1. These Centres of Excellence are responsible for the development, evaluation, and distribution of pain management curriculum resources for health professionals. Are they going to collaborate and produce one unified curriculum? If not, this leaves a big hole open for University-specific bias – eg if one University has a large pain research group, they may choose to include more of their research and leave out potential other work available. If they are not collaborating, they are also at risk of developing multiple resources.. some which will be better/worse than others!
    2. These Centres of Excellence means pain management training and resources may be more accessible for the ‘trainers’, but doesn’t necessarily mean pain management or resources will be more accessible for the PATIENTS.

    My second point comes back to your final point… convenience and consistency play a big part in the success of the treatment. These Centres sound like a great idea from the quick look I’ve done into them, but I’m dubious about how they will change either the conveniency OR the consistency of pain treatment in the US – I’m therefore dubious about how they will influence the success of the current treatments offered.

  4. stuart miller says:

    Thanks Jane for the clarity. Non-threatening visual input seems to be very important, especially early (are there a number of cases of blind patients with CRPS ?). I think that there a few unexplored variables in the success or failure of treatment- being ‘on message’ with the biological rationale is important – going through the motions is ok at first but I think that belief structures are critical (in both the patient and the treatment team and sometimes those outside this circle); consistent and clear feedback that the system will adapt with an appropriate baseline of intervention (if necessary with minimal motor recruitment as with lateralization to start) is important – I would be surprised if the majority of clinicians at those two centres in the Johnson study are ‘on message’ on how to effectively treat their patients. I appreciate the discussion.

  5. Matthias Tomczak says:

    Hello Jane and others,
    i have different kinds of Problems with these recent Review and the studys rely to GMI Treatment. There were Treatment studies with only mirror therapy (Cacchio 2009 etc.) in the review done by you Jane. I do not understand why? Is this GMI or another Kind of Treatment? For me it seems to be another way of Treatment for chronic CRPS. Did I understand lorimer Moseley wrong if he was developing GMI for CRPS Patient in chronic Conditions because of the Study by Candy McCabe (2003) who uses Mirror Therapy alone? I was always asking myself if the Conclusions by Moseley was too early to say that Mirror therapy alone is not helpful in chronic CRPS patients. McCabe had only 4 patients in chronic stage… And CRPS seems to be not very homoguenous in the wide range of different Symptoms we see in patients. Maybe there is no way to seperate them in “chronic” and “acute” but in different symptomatic clinical ways of the desease. And maybe the use of Mirror therapy alone is worth to be evaluated. I know Moseley did a study with variations of the GMI components. But why not mirror therapy in the first place? Or Mirror therapy combined with desensibilisating approaches (as Maihöfner 2004 did but without the visual Feedback of a mirror). I think maybe we go too fast with our Conclusions on the evidence of a therapeutical approach. Something that also been showed by the Johnson study, i think…
    Hope you were able to understand my poor English and please don´t understand me wrong i´m really appreciate your and expecially Lorimers work in this field. Just a few thoughts…

    Jane Reply:

    Hi Matthias,
    Thanks for your great comments!
    Mirror therapy was included in the review because it makes up one component of GMI. I guess one of the reasons we wanted to look at the components separately was to see if it is one component of GMI that makes all the difference. However, our results would suggest that mirror therapy may make the GREATEST difference, but certainly left/right judgements and motor imagery play an important role too (this is also demonstrated by the poor results in the Moseley 2005 paper where patients did GMI in a mixed up order). Certainly mirror therapy seems to be effective in treating some patients, but when combined with left/right judgements and motor imagery in GMI the results are even BETTER.
    I also agree that sometimes we jump to conclusions based on studies done in a controlled way. I think the Johnson study shows that the old saying ‘less is more’ can be important in treating patients. The more therapeutic treatments we try doesn’t necessarily mean the faster they will recover.

    Matthias Reply:

    Hallo Jane,
    thanks for your comment and i hope you had a nice easter break… Thought about your response and still thinking bout it. I mean this study from 2005:

    Is successful rehabilitation of complex regional pain syndrome due to sustained attention to the affected limb? A randomised clinical trial, G Lorimer Moseley (2005) in Pain 114 (1-2) p. 54-61

    when i get everything right were the sequences:
    hand laterality recognition, imagined
    movements, mirror movements (RecImMir, MIP); imagined movements, recognition, imagined movements (ImRecIm); recognition, mirror movements, recognition (RecMirRec)
    so i´m still not understand why not one group with mirror therapy at first place?
    Do we know that it is really not working as a standalone therapy in chronic CRPS? There has to be a bigger trial (n> 60 or something like that) for mirror therapy alone as a treatment for CRPS, don´t you think? How are your clinical experiences? I had patients in the chronic phase who responded very good to mirror therapy alone… Although its not representative (maybe 5-6 patients). Their changes were in the range of clinical importance changes on the VAS and on the PSFS scale. Just case studies…

    Thanks for your time i´m really love to talk about these things with someone who is deep in the subject…

    Jane Reply:

    Hi Matthias,

    You raise a really great point – perhaps there should have also been a group that started with mirror therapy as this certainly seems to have the largest effect as a stand alone treatment.

    I don’t think we can confidently say whether mirror therapy is or isn’t working as a stand alone treatment. You’re right – we definitely need a larger, randomised controlled trial to give us more of an idea.

    It’s interesting to hear your positive outcomes from mirror therapy… have you had anyone with a negative response? I think certainly both mirror therapy and GMI are proving to be effective treatments for some CRPS patients. The future looks bright!


  6. stuart miller says:

    Tara, I wasn’t suggesting it either…it was only brought to mind when I remembered a case involving delayed wound closure with a patient with CRPS Type 1 that only achieved closure after hyperbaric treatment. Delayed wound closure is always challenging. So is working with patients with CRPS. In terms of the most recent review papers, an aberrant inflammatory response (with facilitated neurogenic inflammation), vasomotor dysfunction and maladaptive neuroplasticity (with cortical reorganization) are highlighted as the underlying mechanisms but in terms of clinical practice it is challenging to deal with an extremely swollen extremity with ++ hypersensitivity. It seems with children, that they adapt better to desensitization than others. Tactile discrimination can be done simply with good effect (I have read the study suggested – always looking for something new) and GMI still works (and mirror therapy early with evidence of central sensitization – defining this seems to be a challenge though).

  7. Tara Packham says:

    Stuart, I feel I should clarifiy I wasn’t really suggesting that hyperbaric treatment is a good option for CRPS – it came to mind as an example as it had just come up in an interview I had conducted earlier that week (I am interviewing persons with CRPS about their symptoms and experience as part of a cognitive debriefing study of a self-report questionnaire under development for this population). I was really trying to illustrate the disconnect between etiological theory and current clinical practice.

  8. Hi Stuart,
    Sounds like an interesting case! Yes, one of the clinics did see functional improvements but still with no improvement in pain scores.
    You are definitely right – keeping it fresh is best. That is part of the problem of being a clinician, findings ways to make the leap from research to practice; distance from treatment centres is definitely a barrier to that. For treatments targetting the brain more is better, so seeing patients every 1-2 weeks can be a huge problem. The Johnson article certainly faced the same challenge, I just don’t think they highlighted this important point in their limitations section enough.
    There has been some research in TDT with CRPS – see Lorimer et al.’s paper ‘Tactile discrimination, but not tactile stimulation alone, reduces chronic limb pain’ (2008) here:

  9. You raise an excellent point Stuart – travelling to get treatment. There’s very little intrinsic value of any treatment(s) that if by travelling to obtain them, one hyperstimulates a pain response that negates any benefit (and indeed, can cost vastly more).

  10. stuart miller says:

    I have worked with a patient with CRPS who required hyperbaric treatment for wound closure but not sure if there is any research out there. I had thought that with Johnson study that one clinic (clinic 1) had significant functional change but not pain. I think that in terms of your analogy, with tuna it is important to keep it fresh (the longer it is in the car, the worse it smells). From the studies, the CRPS clinics were located a fair distance from a number of clients so that in many cases they had to do long commutes and would only come infrequently (once every 1 – 2 weeks). I think this was one of the significant variations from the research group. The concept of accessibility and convenience for treatment for a patient with severe pain should be obvious in the context of reducing threat. As well, I know Dr. Herta Flor had done studies on tactile discrimination training with amputees but are there good (?RCT) studies on TDT with CRPS ? Thanks !

  11. Tara Packham says:

    As a clinician and a researcher in the arena of CRPS, I have to agree that this resonates. If we think that these folks have a sensitized nervous system, then we need to consider how much we are ramping up inputs to that system with our interventions. If we think CRPS is inflammation driven, then perhaps we need to evaluate what we are doing to control the inflammatory responses in the tissues. If we think that tissue hypoxia is a contributor, perhaps we should be sending these patients to the hyperbaric chamber…. ? My point is, I think the one of the biggest challenges facing health care practitioners is we are not sure what research we believe about this syndrome (I mean,it took a hundred years just to agree on a name!) and so we don’t hold confidence in any one treatment approach. Furthermore, we still are using treatment approaches (sympathetic nerve blocks, stress loading, desensitization by hyperstimulation) that are based in theories that have been rejected or revised. Knowledge translation is a huge buzzword in academic circles, but its importance is illustrated in this paper. If we fail to “translate” research into clinical practice, it may not be the efficacy of the treatment, but the accuracy of the implementation.

    Jane Reply:

    Well said, Tara!

    And I’m really not sure that our agreement on a name was all that helpful after a hundred years… the name COMPLEX regional PAIN SYNDROME only lends itself to creating more fear!

  12. I’m working with a database on patients with Neuropathic pain and am discouraged with the small effect size of all types of treatment. GMI is not included in this database, but in looking in the literature and anecdotally the effect treatment is modest across the board.
    The only criticism I have with GMI is that it is new enough that the parameters are not entirely clear (especially for the implicit motor imagery). Until clinical practice is able to replicate research protocols I’m not surprised to see differences between RCTS and clinic audits.


    Jane Reply:

    Geoff & Dr Matt,

    I think you both raise a good points. The parameters are not entirely clear for use in the clinic (and let’s be honest, we can’t see patients in the clinic once every couple of days as Lorimer did in his RCTs), so results can fluctuate.

    I suppose my best response to these suggestions is that more research needs to be done. And by someone other than Lorimer. Clinical audits are certainly important to recognise what works and what doesn’t in the clinic, but I think the Johnson paper isn’t reason enough to give up on GMI. It tells us that the parameters they used didn’t work, so we need to persist and work out how we can translate Lorimer’s RCTs to clinical practice.

    What parameters have you used in the clinic Dr Matt? Have you tried different parameters?


    Dr Matt Reply:

    Hi Jane,
    my team had me working with the noigroup recognise program daily for 12+ weeks, starting out simply and working my way into increasingly complex levels as time progressed.

    As I said when banging my head against PhD labwork … “I’ve discovered 10,000 ways it doesn’t work” although perhaps its relevant to add “which led me to the way it did!”

  13. A bold conclusion to the Johnson study there Ms Bowering and one that I find myself agreeing with. Based on the size and variability of the Johnson study group, the conclusions drawn do seem especially courageous.
    Despite my own experience with pain and GMI being one of the “not so effective” category, I can certainly see the potential value of it for others and simply work under the assumption that I’m a statistical outlier.