• lumbar multifidus dysfunction
• back
• pain
• strengthening
• rehabilitation

Regarding the article “Effects of three different training modalities on the cross sectional area of the lumbar multifidus in patients with chronic low back pain”,¹ I would like to extend my appreciation to Dr Danneels and his colleagues for their interest in contributing to the literature on this important and clinically relevant topic. Unfortunately, there are important design and methodological flaws inherent in this study which call into question its results and primary conclusions. I respectfully submit this review of the study, its findings, and the authors' clinical recommendations.

The objective of the investigation of Danneels et al was to determine the potential for different exercise models to reverse the pathology related atrophy of the lumbar multifidus muscle in people with low back pain. As described by various researchers, the lumbar multifidus experiences a number of morphological and neurophysiological changes following low back injury.^2–4 One of these changes is a segmental atrophy which develops at the level of pathology, on the symptomatic side and as quickly as 24 hours after the injury.³ Further, these changes have been shown to persist beyond the resolution of symptoms,⁵ and for at least five years after surgical intervention for intervertebral disc herniation.⁴ There is evidence that such findings are indicative of a neurologically mediated process rather than a simple disuse or weakness phenomenon.⁶

In their study, Danneels and colleagues¹ compared the motor re-education model, originally developed by Richardson et al⁷ and as studied by O'Sullivan and colleagues,⁸ against two variations of a traditional strengthening model. The first of these strength training variations utilised typical concentric and eccentric lumbar extensor loading motions. The other added a static or isometric component which was to be maintained between the concentric and eccentric phases of the exercise.

The authors concluded that, in order to correct the atrophy observed in the lumbar multifidus, patients should perform strengthening exercises targeting the lumbar extensors, ideally incorporating an isometric “pause” into these exercises. Danneels et al reported that this was the only exercise model tested that developed sufficient hypertrophy to correct the multifidus atrophy seen in their experimental population.

These findings conflict with those of Hides and colleagues,^5,9 who have published data showing correction of the pathology-induced lumbar multifidus atrophy using a considerably more specific and subtle activation of the multifidus muscle.⁷

Unfortunately, the method by which the authors measured the cross sectional area (CSA) of the multifidus muscle introduced a large degree of methodological error, calling into question the study's findings and therefore its clinical recommendations. Using computed tomography scanning, Danneels et al took measurements from three arbitrary levels of the lumbar spine (the L3 superior end plate, the L4 superior end plate, and the L4 inferior end plate) recording the CSA of the multifidus muscle at each of these levels bilaterally. They then summed the right and left multifidus CSA at each of these segments resulting in a single multifidus score for each level.

In preselecting the levels from which CSA measurements would be taken, it is possible, indeed even likely, that the pathological level would have been missed entirely in at least some of the subjects. This is important because previous studies have shown that the multifidus muscle experiences its greatest loss of CSA at the primary level of pathology.^3,10 Also problematic was the decision to sum the CSA scores at each of the preselected levels. A number of studies have shown that the lumbar multifidus, ipsilateral to the pathological side, experiences a cascade of neuromorphological changes, including atrophy, in the presence of both acute and chronic pathology, whereas the multifidus contralateral to the pathological side experiences no such changes.^2–5,10

Taken together, the preselection of measurement levels and the summation of the bilateral multifidus CSA would have had the effect of attenuating any side to side differences in multifidus CSA even if a subject's level of pathology happened to coincide with one of the levels from which measurements were obtained. Through either of these mechanisms, this measurement scheme would have introduced a substantial mass of healthy, non-motor dysfunctional muscle into each of the multifidus CSA measurements. The net effect of the measurement approach employed by Danneels et al, although probably intended to more discretely reflect the pathological side multifidus, was to bias the outcome of the experimental intervention toward an exercise model with the ability to cause hypertrophy in healthy muscle—that is, a strengthening exercise. It is likely that the “dynamic-static” exercise recommended by the authors as being most effective for correcting the pathological atrophy of the multifidus instead caused hypertrophy of the non-pathological, non-atrophied multifidus muscle segments included in the three measurement scores. Ultimately, the study's recommendations are unsupportable given this flaw in methodology.

The low load multifidus activation exercise, developed by Richardson et al⁷ and used by O'Sullivan and colleagues,⁸ is to be performed as a co-contraction with the transversus abdominis muscle, and is intended to correct a neurologically mediated loss of normal multifidus muscle volume, not unlike that seen in the vastus medialis following trauma or surgery involving the knee joint.¹¹ In studies in which the CSA of a pathological multifidus muscle has been compared with its contralateral and “healthy” segmental partner, this form of motor re-education exercise has been shown to normalise the CSA of the pathological multifidus in as little as four weeks.⁵

It is critical that both researchers and clinicians appreciate that a significant body of research now shows that the “atrophy” seen in the multifidus muscle in people with low back dysfunction is representative of a form of impaired motor control, not simple disuse weakness. As such, traditional strengthening exercises will often fail to correct this fault, just as daily physical activities fail to maintain a normal segmental CSA at the pathological level. Certainly, the historical lack of success of the rehabilitation and medical professions in treating low back pain using the wide variety of strength based clinical models used over the last 50 years should serve as sufficient motivation to look to more evidence based models as an explanation for the condition. The motor control dysfunction model as developed over the past decade by a variety of researchers^{4,6,7,9,12,13} holds great promise, both as a basis for understanding the causes of back pain and in developing more effective treatment strategies for our patients.