Confounding and the Biopsychosocial Model
As I blogged a few days ago, the biopsychosocialists are back. We dealt with these theories in pain management and orthopedic injuries a decade or more ago, but advances in pain research made these theories go away or at least fade into the background. Now, they have again surfaced in a new physical therapy viral trend that’s frankly disturbing.
BIOpsychosocial or bioPSYCHOSOCIAL?
The idea put forth by the biopsychosocialist belief is that patients with chronic pain have a biologic component that interacts with psychology and their social context. As a basic idea, it’s innocent enough. We’ve seen many patients through the years who need to be educated about what’s wrong when they’ve developed the wrong idea about their disease. For example, they believe that’s what’s on their MRI is awful, but in fact, it may be little different than the MRI of most people their age who don’t have pain.
However, in practical experience, through the years the biopsychosocialists have always meandered off the proverbial farm. Rather than considering pain a physical phenomenon that can sometimes impact psychology (BIOpsychosocial or Bps), or that tissue damage and psychology are equally weighted (BIOPSYCHOSOCIAL or BPS), they almost always trend towards considering psychology dominant in this calculus (bioPSYCHOSOCIAL or bPS). As an example, the leader of the movement, a PT named Pete O-Sullivan, recently tweeted that the SI joint is bombproof. With that, like all of the biopsychosocialists I’ve known in the past, he was demonstrating the point. The SI Joint is a structure that many studies have shown gets injured and is not injury proof, but the temptation to turn Bps or BPS into bPS is just too great.
Why? Treating patients with chronic pain is very, very difficult. It takes many years to understand which end is up and which therapies will likely help which patients. This takes time, energy, resources, education, and the desire to get constantly better at your art. However, pointing the finger at the patient is just easier. You don’t need to become any better at treating patients because the reason the patient isn’t improving isn’t due to the clinician’s lack of skills, it’s the patient’s psychosocial belief system. Hence, the bioPSYCHOSOCIAL (bPS) model allows clinicians an easy cop out.
Confounding and the bPS Model
Bad research into the bPS belief is deceptively easy to perform. Why? You simply administer a questionnaire that measures catastrophizing and correlate that to the outcome of a therapy. Catastrophizing means that the patient believes that their illness is worse than it really is and as a result, they believe they won’t improve or get better. So if higher levels of catastrophizing correlate to worse outcomes, you have a nice little study. The problem is that there are two reasons patients have these beliefs. One is that they have a misconception about their injury provided by their providers and the other is that they are legitimately injured and based on reasonable observation of their own condition, reasonably hold the belief that they won’t improve.
Confounding is a real problem in research and is king in bPS studies. Confounding is when something that hasn’t been understood may be causing the effect you observe in your research. As an example, if you notice that a group of people have higher anxiety than the normal population when they work, you might conclude the group has an anxiety disorder. However, if you didn’t note that the group you studied was made up of mostly test pilots, then your conclusion is erroneous, as it’s reasonable to conclude that test pilots have an anxiety provoking job. The same is true in bPS studies. Chronic pain due to tissue injury is anxiety provoking and if your experience is that you’re not improving, it’s reasonable to catastrophize. The problem for almost all of the bPS studies ever performed is that we have no good way to quantify tissue injury.
How Did We Get Here? Imaging Pain with a Structural MRI is a Bad Idea
The bPS model began after it became clear that MRI didn’t explain many chronic pain conditions. Given that an MRI shows the structure of tissue and oftentimes there was a poor correlation between that structure and pain, the conclusion was that the pain wasn’t due to tissue damage, but instead due to other factors like psychosocial issues. However, that boat began to take on water with the following studies:
- Research by Apkarian on how parts of the brain atrophied in patients with chronic low back pain and that atrophy had no correlation with bPS factors
- A study showing that when you were able to kill the pain signals coming from injured facet joints in the neck in whiplash patients, their anxiety and other psychosocial metrics normalized
- Studies showing that the multifidus muscle in the low back atrophied in patients who have chronic low back pain
In addition, there have been a multitude of recent studies showing very real changes in the structure, neural activity, and chemistry of the brain in patients with chronic back pain (see ref 1, ref 2, ref 3, ref 4, ref 5, ref 6, ref 7, ref 8, ref 9, ref 10). We also now have research on chemical biomarkers for chronic pain that are now being interpreted using machine learning to better understand how changes in the body’s patterns of cytokines, growth factors, and hormones link to pain (ref 11). I note that all of these papers were just published in the last 18 months!
Any of the papers published during that time that tried to link brain changes to the bPS model used questionnaires to detect pain exaggeration that were developed before we knew that chronic pain was physical. Meaning they all lacked an internal control to validate the questionnaire itself against any these new technologies that can classify and quantify pain. Hence, like much bPS research, they intellectually chase their own tail.
bPS Studies Have the Same Weakness
While we can now image pain, the bPS research ignores these new technologies and instead focuses on the fact that simple MRI scans often fail to correlate with pain. Given that every bPS study would require 100% certainty that any given patient wasn’t actually injured or only minimally injured, this is a real problem. After all, you need to eliminate the possibility of real tissue injury in order to say that anxiety, fear of movement, and catastrophizing isn’t reasonable. Given that these new technologies above have shown that we can now image pain or it’s aftermath, any bPS study worth the paper on which it’s written (or the bytes it takes up on a hard drive) would need to begin with these advanced imaging scans to place patients into a low injury category. If the study doesn’t do this, it risks being confounded by the responses from patients who are legitimately and seriously injured and who justifiably aren’t right in the head!
The NHS has Helped Spread the bPS Model
National health systems like those in the UK and Canada have always had a US problem. Our well-monied health system allows for some of the most technologically advanced care on earth en masse. Add in funding from NIH that dwarfs anything these governments can muster and you get a health care arms race. The citizens of these countries want all of the glitzy stuff available in American healthcare, but paying for these technologies in a government funded system is tough. Hence, they have no choice but to make cuts elsewhere and patients in chronic pain are easy pickings, as they often have no way to substantiate their diagnoses with structural MRI imaging. As a result, bPS proponents have always done well in these countries. They are making the downhill argument to underfunded national health systems when they tell politicians that chronic pain patients have nothing really wrong that can’t be fixed with a good psychology session or re-education. Hence, it’s not surprising that these changes denying disability payments due to bPS philosophy in the U.K’s NHS have recently come under fire from academics.
My biggest concern is that we’re now teaching a generation of physical therapists in the UK and Canada (and more recently the US) to take the easy way out. Rather than learning a craft and spending the thousands of hours they need to get educated about how to help patients in chronic pain, we’re letting them take the bPS cop out. When the patient doesn’t get better, it’s the patient’s fault and not the clinician’s.
As a clinician who has worked with chronic pain patients for 25 years and teaches fellows in this area, I always tell them that they don’t know enough to be able to declare a poor result the fault of the patient’s psychology. My mantra has always been, if the patient doesn’t improve, you missed something. If he or she doesn’t improve after you went back to the drawing board, you still missed something. Only after that attempt can you consider that there might be a sliver of truth that this patient may have bPS issues.
This key process of self-criticism isn’t happening in the bPS camp. Yet only through this process can clinicians who treat patients improve their skills. What would happen for example if a ball player who didn’t get the ball where he or she wanted to go always blamed the ball rather than their own lack of skill?
The upshot? Just when the bPS model is dying in a deluge of research showing that chronic pain is quite real and can be objectively measured, a physical therapy viral phenomenon is arguing that the spine is a bombproof structure that doesn’t really get injured. This is bizarre, as we have little evidence that any of this is true and more concerning is that it’s an insult to legitimately injured patients everywhere. Even more concerning is that the bPS gang is teaching a new generation of clinicians to bypass a much-needed mechanism of self-evaluation of their own clinical skills that’s critical for learning how to help difficult patients.