A complaint commonly heard from the researchers of physical therapy is that clinical instructors frequently make claims for which there is no evidence. Often there is some truth to this but more usually the term evidence is being used in a form that is convenient for the complainant. Partly from this complaint comes the perceived need for evidence based practice. On the surface this seems a reasonable requirement for all practice but the term “evidence based” needs to inspected in detail before clinicians give up 90% of their practice because there is no criterion or other high level evidence for the practice. First the clinician is not to blame fault for the paucity of high evidence supporting their practice. It is the job of the researchers to prove practice and the fact that there is so little validation of our practice together with good sensitivity and specificity numbers for our tests or experimental proof for out treatments is an indictment of how poorly our clinical-research teaming works in North America as compared with Australia. It not that those tests or treatments used by clinicians are not valid, but that they are neither validated nor invalidated. In these cases (the majority) other forms of evidence must be used. This evidence may take the form of construct, consensus, case studies, even opinion and while these forms may not be the strongest evidence there is they are frequently the only evidence there is. If we lived in a perfect world, the clinicians would go to the researchers with a question that followed an observation and the researchers would do their best to answer it. The clinician would then take the research and incorporate it into their practice. This not being the perfect world generally the clinician, especially the newer graduate must go to the experienced clinician and the instructors in manual therapy for the best evidence available. It is then that we run into the problem we started with, claims to truth that are not only unsubstantiated by research but more frequently have never been researched in the first place. The clinician is now at fault for stating as a truth what may be backed by lower levels of evidence but is rarely the truth. Even when there is experimental proof of a test or treatment this merely states a fact and not the truth. Truth in the philosophical sense and fact are not necessarily the same. The best examples of this dichotomy comes from the hardest of sciences, physics. Relativity and quantum mechanics are the two most proven theories extant, neither has failed to agree with observation and both have been proven to about 12 decimal places. Yet the underlying “truth” to each theory is not known. Physicists use models that incorporate the facts of these theories to form a framework that allows them to visualize the universe that allows them the best conditions to do their work. For quantum theories there are many models of how the universe work at a sub-atomic level none of which are of necessity compatible with any of the others. The point of these models is not to tell the truth but to explain the facts in an organized and rational manner. A good model incorporates all or most of the facts, a weaker model less facts and a downright bad model no facts and should properly not be called a model but fantasy. When a model fails to incorporate new facts as they are discovered then it is modified or replaced until the new version does so explain them.
Instructors of manual therapy would do well to follow this example and make it clear that what they are describing is a model and not the truth. This way we do not look silly and more importantly the specialty does not loose credibility when our “truths” turn out to be a lie. One example of this in manual therapy can be seen with the explanation of how manipulation works (whether it does or not can be debated at another time). The two main explanations or models are the mechanical and the neurophysiological. The former simply holds that manipulation is effective by moving a joint that is fixed (and the exact mechanism of fixation is a variable part of that model) at one end of its range and eliminating the abnormal stress put upon a pain sensitive tissue. The neurophysiological model explains manipulation’s efficacy by having the sudden mechanoreceptor input into the spinal cord segment and more central regions cause pain modulation and muscle tone relaxation and so decrease or eliminate pain and increase range of motion. In all probability there is an element of “truth” in both of the models but providing that the user understands that the adopted model is simply a working framework that helps visualize what is going on it does not really matter which model is chosen. You pays your money and you makes your choice and the truth may never be known.
A good model is not necessarily the truth, in fact it usually is not the truth nor anything like it, but rather it is a conceptual framework that allows its user to visualize the phenomenon and so more effectively understand the relationships between the facts of the phenomenon and between the phenomenon and the observer. In manual therapy it allows us to bundle the examination clinical findings into a diagnosis and treatment. The best model incorporates the most facts or observations and the worse model the least. As new facts emerge they are either successfully incorporated into the model (that is explained by it) and the model becomes stronger or, if the new facts cannot be incorporated, the model is modified or abandoned and a new one substituted. Failure to expand, reduce or abandon models as more information becomes available or treating concepts as truth results in credibility loss and/or confusion when conceptualizing the phenomenon.
There are two other conditions for a good mode in addition to incorporating as many facts as possible; it is economical and it drives an intervention whose result is predictable. For the first condition, in concordance with the principle of Occam’s razor, the model should be as simple as the facts allow and for the second condition, the interpretation of the model should, for us at least, lead to a test or treatment whose outcome can be predicted from the model. During the reading of this article remember that no model is the correct model as correct is not a term that should be applied to models, you simply have to decide if it fits the clinical facts and that the model works for you.
Generally ideas on biomechanical dysfunction of the spinal segment fall into one of three main camps, segmental instability, segmental hypomobility and a mixture of the two. Currently segmental instability is enjoying a good deal of popularity thanks mostly to the research coming from Australia particularly from the research group headed by Jull, Richardson, Hides and Hodges1. It is worth noting that although these people have done some excellent work on biomechanical instability they did not invent the concept. This has been around for many years but did not have the respectability of good research nor did it really have a good physiological underpinning. Unfortunately, as is the usual case, moderation goes out of the window in the light of good research and now many therapists have thrown out the idea of hypomobility as a major consideration in mechanical spinal pain. I hasten to add that this is not the fault of the researchers in the field but rather of the need of the consumers of the literature to need something new and fashionable.
Hypomobility on the other hand had been overly-dominant for too long with mobilization and manipulation being almost the only techniques used to treat segmental biomechanical dysfunction with the odd, non-specific exercise thrown in to differentiate us from the chiropractors, who our legends have it, do not give exercises (how easily an entire profession can be dismissed). A mixture of the two ideas has been around for a long time with manual treatments still taking the pre-eminent place but with a nod to the need for an assessment of and treatment for instability if found. So for the purposes of this discussion we can take a look at the two pure theories of segmental dysfunction and see how they fit the facts as we know them.
Instability is currently and generally considered the result of failing segmental stabilizing muscles, mainly the multifidus, transverse abdominus, and maybe psoas in the lumbar spine, the multifidus and the prevertebral muscles in the cervical spine and multifidus, piriformis and the pelvic floor muscles in the pelvis plus the larger muscles such as latissimus dorsi and the adominal obliques. (By the way if you ever get the opportunity read McConnaill’s work on shunt muscles, this is perhaps the earliest reference by far to the role of muscles as stabilizers.)
Pain or reflex inhibition causes the muscles that control translation to become inadequate and there is an increase in the neutral zone resulting in more reflex inhibition that then leads to end zone instability at which time pain and dysfunction may make an appearance. Just to complete the picture, or at least add to it, it would seem likely that not all insidious segmental instability is the result of muscle failure but may result from excessive mechanical stress applied over a prolonged period of time. The observation that hypomobility of the hip is associated with low back pain would tend to support this. The consensus among clinicians that have written on the subject is that the patient suffering from segmental instability will present with most if not all of the following2,3,4,5.
- Short duration episodic pain
- Minor triggers causing the pain
- Mild to moderate pain
- Mild to moderate referral
- Full but abnormal range of motion between painful episodes
- Abnormal segmental end feels
- Treatment fail to provide long term relief
These observations plus the research using EMG, MRI and biopsy studies are the “facts” of the model of segmental dysfunction being caused by segmental instability. The model proposes that mechanical spinal pain is the result of instability but the actual mechanism for pain production is very vague and is often cited as muscle spasm or hypertonicity causing painful ischemia of the muscle or the loss of control of the segmental allowing slipping to result in inflammation. The success of the model comes from the incorporation of these “facts” and therapist’s acceptance of the model. Episodes of pain are explainable by minor trauma nudging the unstable segment into a painful state, the abnormal movements during the pain free phases are explained by lack of normal motor control, The short duration of the episode by minor trauma nudging it back again and the fact that treatment fails to provide long term relief by it not adequately addressing the motor control of the segment. It is economical in that it proposes only one pathology to explain the facts rather than a different pathology for each fact. Its weakness lays in not explaining why mobilization or manipulative therapy is so successful in reducing the patient’s symptoms and increasing range of motion sometimes immediately and the method of pain production. If the pain is the result of increased tone or spasm in the muscles then this is at odds with the idea of inhibition being the root of the instability. If the pain is caused by inflammation why can this pain be eliminated by mobilization or manipulation as this obviously cannot beneficially affect the inflammation.
The hypomobility model incorporates the same observational facts as the instability model but does not have the same research backing it up. In this model the patient’s pain and dysfunction results from a joint that is jammed or subluxed or fixated or whatever term is current and local. In essence the joint is caught at one end of its range and cannot move in the opposite direction. For example, a joint that is jammed into extension would cause a flexion hypomobility affecting both the passive physiological movement and the arthrokinematic associated with that movement. The mechanism of the subluxation is itself a sub-model and ranges from incongruencies in the secondary or tertiary contours of the joint, through meniscoid entrapment to misalignment of the primary contours. There is also a neurophysiological sub-model to explain this jamming which consists of hypertonicity of the muscles from pain (a somewhat twisted argument as the pain causes the hypertonicity and the hypertonicity causes the pain). Apart from the lack of research evidence, which in itself is not a fatal flaw in the model, the failings are that in the absence of overt trauma there is no good mechanism for the subluxation to occur and there is no good reason for recurrences to occur, that is for the problem to be short term episodic.
The mixed model is better in that address all aspects of the “facts” and integrates them. The mixed model goes something like this. There is an underlying basic unstable state or matrix, which is in itself painless. This explains the painfree episodes. But a minor force or trigger subluxes the joint which may then cause a painful phase (however not all subluxations are painful as we can attest from examining asymptomatic students on courses). The pain is caused either by overstretching of the ligaments and capsule of the subluxed joint or by adjacent joints becoming painfully hypermobilized as they are overstressed by the subluxed joint. This explains
subluxed and so the condition is vulnerable to mobilization or manipulation. The patient improves, often becoming completely painfree with full movements restored immediately but fails to maintain the improvement for varying periods depending on how unstable the segment is or what activities the segment has to tolerate. These last considerations explain why not all instabilities are painful or dysfunctional and why a previously asymptomatic person can become chronically painful with say two or three jogging sessions even though the running is discontinued quickly. It also explains why whiplash can have such a devastating effect on a previously asymptomatic subject even though the delta-v is so low. Consequently the mixed model is is better than either of the two pure models in that it incorporates both sets of clinical observations, the research findings on segmental instability and the effectiveness of manual treatments in reducing or eliminating the pain. It also explains the episodic nature of the condition that the hypomobility model fails to do and the pain mechanism that is poorly explained by the instability model. Lastly it is economical in that to properly explain the dichotomous nature of the mechanical spinal pain sydrome each pure model would have to propose two or more pathologies, one for the painful phase and the other for the pain free phase.
Finally a name should be given to the mixed model that describes the model adequately. I propose the term Phase Transition Model. A phase transition is where there is a radical change in the behavior characteristics (properties) of s a substance or a system without a proportional change in the composition of the substance or system6. The term is used in physics, particularly cosmology where it is used to describe the big bang and the inflationary period of cosmic evolution but a simpler example is the phase transition that occurs when water changes to ice or vice versa. Water and ice are almost two completely different substances as far as their behavior is concerned. Ice cannot put out fire until it changes to water, ice is extremely slippery, water considerably less so, a bucket of water tipped over you from 15 feet will make you wet, a bucket ice will make you dead. While the term is strictly used to describe fundamental property changes in thermodynamic systems there is no reason why it should not be used here with good effect. The phase of instability is painless, mobile and leaves the patient fully functional. Assessment during the unstable phase with stability tests will, on a good day, demonstrate instability and the passive physiological tests hypermobility while during the painful phase passive physiological mobility test will demonstrate hypomobility with all the characteristics of subluxation and the stability tests will show stability. In effect the segment goes from the matrix of instability to hyperstability and from a painless state to a painful state. It undergoes transition from one phase to another and back again as effective forces are applied.
In summary this article has looked at the construction of a model to explain the clinical observations and research findings of mechanical spinal pain. Two pure models were discussed and found to be inadequate explaining all of these observations and findings.
A mixed model which incorporates both pure models was found to be a better model in that it explained all of the clinical observations as well as incorporating the research in an economical manner. I have suggested that the term phase transition be used to name the mixed model. The advantage to this type of approach is that the user knows going in that it is not the truth and so when new facts emerge that are not explained by the model it can be abandoned or modified with no credibility loss for the profession or individual and no sense of loss as there is no personal investment in the model as there would be if the author really believes that he or she knows the truth. The truth may in fact never be known and it certainly is not know currently but with a rational and scientific approach the condition of mechanical spinal pain may be usefully visualized and that can only help in its diagnosis, prognosis and treatment.
- Richardson, C. et al. Therapeutic Exercise for Spinal Segmental Stabilization in Low Back Pain. Churchill Livingston, Edinburgh. 1999
- Grieve, GP. Lumbar instability. Physiotherapy. 68:2 1982
- Lee, D. The Pelvic Girdle, 2nd Edition, Churchill Livingston, Edinburgh. 1999
- Meadows, J. Orthopedic Differential Diagnosis in Physical Therapy. McGraw-Hill, NY. 1999
- Schnieder, G. Lumbar Instability. In Grieve’s Modern Manual Therapy. 2nd. Edition.
- Boyling, JD. Palastanga, N. Churchill Livingston, Edinburgh. 1994