Was pleased to receive an email from Demetry Assimakopoulos a few days ago. Demetry is a Chiropractor with the Toronto Rehabilitation Institute's Comprehensive Integrated Pain Program. His question was good and since I suspect others may have a similar question, I requested (and received) his permission to share the exchange below. I hope others may find it helpful. If you want to read the paper to which he's referring you can do so here and you can get your own blank copy of the pain phenotype radar plot here. And, if you're so inclined, why not keep an eye on the Handspring Publishers website for a book release on this topic from me and Dr. Jim Elliott in 2019.
Good day, Dr, Walton:
My name is Demetry Assimakopoulos. I am a chiropractor working in the Comprehensive Integrated Pain Program at the Toronto Rehabilitation Institute. I came across your article in Musculoskeletal Science and Practice, and was excited, as I am an advocate of sub-categorization of chronic pain complaints into dominant etiologies. I was especially intrigued with your "Central Nociplastic Change" category. I have made similar arguments for years.
In your article, you advocate for grading the severity of these separate pain etiologies/subcategories. Much of the literature has identified ways for clinicians to identify these potential pain etiologies using various historical, physical examination and psychometrics/questionnaires. However, the body of literature has not identified any method of grading severity of these, but rather only allows us to identify them as a potential etiological factor. For instance, the PHQ-9 may enable us to stratify the severity of depression, but not the strength of the relationship of depression to pain.
I wonder what methodology you use to stratify the "severity" (for lack of a better word) of these specific pain etiologies. How can one truly, precisely and clinically delineate the difference between someone that falls into your "high" or "very high"categories, for instance?
I ask the above mentioned question, as clinically, these pain sub-categorizations are looked at in a binary fashion (ie. either they are a factor, or they are not).
Hi Demetry, thanks for reaching out and glad to hear you found the paper interesting. Thank you also for the question, great to hear from others who are thinking critically about these things.
In response to your question "How can one truly, precisely and clinically delineate the difference between someone that falls into your "high" or "very high"categories, for instance?" the simple answer is, as of today, we can't. That's actually why we've quite intentionally left it as a qualitative grading system for now, despite some reviewers wanting us to put numbers on them. As you quite rightly point out, the evidence is simply not there yet, not even in a unidimensional way let alone understanding the complex interactions that no doubt exist. However, we do sincerely believe that the concept of triangulation can at least provide a starting point for people to estimate a patient's location on these qualitative domains. For example, if someone were to appear to hold rational beliefs about their pain, show no strong evidence of psychopathology (whether clinically diagnosed or subclinical), demonstrate effective DNIC on some kind of conditioned pain modulatory evaluation, endorse a strong social support network, and on clinical observation demonstrate local tenderness and a consistent pattern of symptom reproduction that appears to match well with what we know of spinal biomechanics, then we'd likely place 'very high' confidence in a primary nociceptive/physiological driver for their pain experience. On the other hand, if they describe feelings of hopelessness for their future, they score over-threshold on a PHQ-9 or BDI (or related tool), speak in terms of withdrawing from their social network, and show no consistent pattern of movement-based symptom reproduction, we'd likely nominate low confidence in a primary nociceptive driver but high/very high confidence in an emotional driver. The primary difference between this approach to clinical phenotyping and traditional subgrouping is that it also allows for a patient to have both an emotional and nociceptive driver (and other drivers as well). If done properly it should at the very minimum facilitate clinical decision making and, we believe, improve patient buy-in.
You're probably picking up by now that this is really founded on the concept of pattern recognition. We use the term 'triangulation' as a sort of guide, reminding clinicians that they shouldn't make any statement of confidence based on a single data source, or even two, but that three sources that all point in the same direction should be adequate to let you make a call on each of the domains (if it walks like a duck, quacks like a duck, and looks like a duck, good chance that it's a duck). Of course, if only 2 out of 3 point in the same direction, then you adjust your confidence in each domain accordingly.
You're also likely picking up that that we're currently using the concept of confidence rather than diagnostic accuracy. If you're familiar with the GRADE system for systematic reviews (a system that helps researchers establish a qualitative level of confidence in the current state and direction of knowledge in a field after considering all available information) that would be a good analogy. Drawing that analogy slightly further, this is more akin to a systematic review approach than it is a statistical meta-analysis approach.
All that said, we also believe that this framework is likely to serve as a 'transitional' stepping stone between traditional approaches of creating distinct subgroups (which rarely work in practice) and what we are working towards now which will be big data-based analytics for deep phenotyping of patients. The radar plot and triangulation is a means to get people thinking in a slightly different framework, but we believe it's far from the end of the story.
Hope that helps, and feel free to reach out any time.