A little while ago now*, O’Keeffe et al published a systematic review and meta-analysis that showed little difference in effect between treatments they described as physical, psychological or combined. The paper was vigorously criticised by Robert Gatchel—of functional restoration fame—and John Licciardone, who run a combined all-on-one-site interdisciplinary program. It was strongly worded – arguing that the paper by O’Keeffe et al sets the pain field back by undermining the great advances in gaining traction for a biopsychosocial model of pain.
I was interested in the systematic review and meta-analysis before I saw the critique because I thought the study was well conducted and asked a potentially interesting question. For some background, here are the broad definitions of the categories they used:
Physical interventions: aim to enhance physical capacity by using methods such as exercise, manual therapy, and ergonomics
Behavioral/psychologically informed interventions: use educational, cognitive, or psychological strategies to enhance behaviors, cognitions, or moods
Combined interventions: aim to target physical and behavioral/psychological factors contributing to a patients’ pain
Here is a short synopsis that Mary and Kieran prepared for BIM, with a few words from me in there too…:
Criticism for following Recommended Practice
We published a systematic review and meta-analysis comparing the effectiveness of conservative interventions for non-specific chronic spinal pain (NSCSP). The review was detailed in a previous Body in Mind post. The review found that current interventions (physical, behavioural, combined) for NSCSP have small similar effects.
[LM: It is probably worth observing that the effect sizes of all three categories are small.]
A recent letter to the editor criticised multiple aspects of our review methodology: using a random effects model, halving of sample sizes to avoid double counting of participants, assessing heterogeneity using the I2 statistic, lack of consideration for a network meta- analysis, no supplementary documentation detailing our analysis.
[LM: My impression of the main criticisms were (i) that Mary’s team didn’t understand the biopsychosocial model and that their study failed to support what Gatchel and Licciardone know to be true – that interdisciplinary pain management is better than either of the other approaches, and (ii) that Mary’s study is politically naive and destructive.]
In our reply, we highlighted our use of widely accepted Cochrane review methods for performing our systematic review, and felt the feedback therefore applied to all reviews conducted using the Cochrane approach rather than our review uniquely. We acknowledged that current methods of meta-analysis may be suboptimal, and perspectives on what constitutes optimal methods for conducting and analysing research evolve. For example, the value of a “p < 0.05” cut-off for statistical signiﬁcance is increasingly challenged. Critically, the author did not provide validated alternative approaches. Most importantly, the author of the letter acknowledged that his suggested changes would likely not change the results, or conclusions, of the review.
Letters to the editor can be very useful – we have written a few ourselves – as a means of post-publication peer-review and commentary. Even in this case, the letter highlights aspects of meta-analysis that could be enhanced. However, criticisms that do not offer better alternatives are of limited value.
[LM: I am pretty sympathetic to the criticism with regard to threatening the credibility of the BPS model. I don’t actually think Mary’s paper threatens it—to even encourage physical activity in people with persistent pain implies some adherence to the BPS idea, because it is clearly suggesting that pain is not a clear indication of structural pathology—if it were, we would not be recommending exercise or indeed mechanical loading of the tissues. That is, I suspect the BPS model penetrates all three categories to some extent.
Even if they weren’t and the evidence showed that ‘combined treatments’ based on the BPS model were ineffective, it would not undermine the model – it would just show the treatments aren’t effective.
I think there is still great confusion ‘out there’ between these two things – BPS model of pain and BPS model of pain-related disability. I think pain itself is truly BPS, and pain-related disability is too, but I think the latter is way more endorsed than the former, even among people running combined BPS-based pain programs.
How do we assess whether we think pain itself is BPS or not? To start, how easily do these phrases roll off your tongue: pain is detected by the brain; central sensitisation drives pain messages in the dorsal horn; central sensitisation means that pain is unavoidable but suffering is optional; sometimes you feel the pain, other times you don’t; etc etc. I don’t think any of these statements can be defended on the basis of contemporary pain science.
Enough from me. Do yourself a favour and check out the full paper, full critique and full response from the original authors. I fully support the debate and dip my lid** to all of them for engaging in it.
*As Chief Editor of BodyInMind, I have two declarations related to this post. First, I think pain is biopsychosocial in nature. I think contemporary models of brain function predict the biopsychosocial model of pain—the pain emerges in response to a mix of cues that can come from the bio, psycho or social domain. I think the body of evidence from human experiments, experiences, cross-sectional and clinical trial studies, clearly support my position. My research group undertakes research into the role of the brain and mind in the development and treatment of persistent pain. Second, the authors of the text below responded to my request for a short sharp summary of their systematic review very quickly, when I asked them back in February, but I have been very tardy indeed in getting this out there. So, my public acknowledgement of tardiness and apologies to the authors.
** ‘take my hat off’]
About the Authors
Lorimer is Foundation Chair in Physiotherapy and Professor of Clinical Neurosciences at the University of South Australia, and Senior Principal Research Fellow at Neuroscience Research Australia. He has published 300 scholarly works. He leads the Body in Mind Research Group, which investigates the role of the brain and mind in chronic pain. For full bio, go here.
Dr Mary O’Keeffe
Mary is an Irish physiotherapist and early career researcher. She is currently on a Marie Skłodowska-Curie Postdoctoral Fellowship in Professor Chris Maher’s research group in The University of Sydney (Australia). Mary’s fellowship involves two years in Sydney, a three-month secondment to EFIC in Brussels (Belgium), and nine months in the University of Limerick (Ireland). Mary completed her Ph.D. in 2017 in the University of Limerick (Ireland). Her Ph.D. focussed on the effectiveness of individualized multidimensional care for chronic low back pain.
Dr Kieran O’Sullivan
Dr. Kieran O’Sullivan is a specialist musculoskeletal physiotherapist. He is a senior lecturer in the Department of Allied Health at the University of Limerick in Ireland. Since 2016, he has taken a career break to lead the development of a new Spinal Pain Centre of Excellence at Aspetar Orthopaedic and Sports Medicine Hospital, in Doha, Qatar. His research group disseminate research via www.pain-ed.com
Mary O’Keeffe, Helen Purtill, Norelee Kennedy, Mairead Conneely, John Hurley, Peter O’Sullivan, Wim Dankaerts, Kieran O’Sullivan (2016). Comparative Effectiveness of Conservative Interventions for Nonspecific Chronic Spinal Pain: Physical, Behavioral/Psychologically Informed, or Combined? A Systematic Review and Meta-Analysis. The Journal of Pain 17(7), 755-774
David C. Hoaglin (2018). Problems in Meta-Analysis of Comparative Effectiveness of Conservative Interventions for Nonspecific Chronic Spinal Pain. The Journal of Pain 19(2), 228–229
Mary O’Keeffe,Kieran O’Sullivan (2018). Criticism for Following Recommended Practice. The Journal of Pain 19(2), 230–231