Promising results from knee osteoarthritis study


A team including academics from the University of Salford, the University of Manchester and Tim Cacciatore from University College London has published a research report entitled ‘Reductions in co-contraction following neuromuscular re-education in people with knee osteoarthritis’[1] that suggests that 20 lessons in Alexander Technique could have significant benefits for people with the condition.  

This was a pilot study with 21 participants, all with an X-ray diagnosis of knee OA, recruited from GPs across Manchester.  The Alexander intervention involved two one-to-one lessons a week for eight weeks, followed by one lesson a week for four weeks, making the protocol similar in scale to that found in studies such as the ATLAS trial. None of the participants had had previous AT experience.  Only one of the original 22 recruits dropped out – after ten lessons – and all lessons were delivered by a single teacher.  There was no clinical control group, but biomechanical data were collected from 20 healthy subjects for reasons given below; this group was closely matched on key variables such as male/female, age, Body Mass Index, height and weight.

Amongst the selection criteria for inclusion was pain when walking. The participants are reported as having ‘relatively high’ pain scores at baseline; 15 of them were taking analgesia for knee pain.  Similarly, the radiographic studies of the degree of osteoarthritis amongst the participants using the Kellgren and Lawrence scale (maximum possible value=4) are reported to show 11 participants classified with KL=2, 9 with KL=3 and four with KL=4.[2]

Funding and delivery

The study received a research grant of £23,000 from the Bupa Foundation (not to be confused with the Bupa UK Foundation - see comment following this article).  All the lessons were delivered by Peter Bloch, MSTAT.

Clinical outcomes

The primary clinical outcome of the study was a self-reported ‘WOMAC[3] pain score’ (a subset of 5 items from the full WOMAC inventory[4]); the secondary outcome was the full WOMAC score.  Importantly, these outcomes were measured not only directly after completion of the Alexander Technique lessons but also at 15 months (i.e. 12 months after the end of the intervention), with six participants lost to the final follow-up.

The results were promising, with a mean reduction of 56% in the WOMAC pain score and of 54% in the full WOMAC score immediately after the intervention and reductions of 51% and 43% respectively at 15 months (p<0.01 in all cases).  Of those using analgesics, 10 had reduced or stopped using them and 5 had maintained the same level immediately after the study.[5]

Bio-mechanical aspects

Experimental set-up

As indicated by the title of the research report, the study sought to explore the possibility that raised levels of co-contraction in knee extensor and flexor muscles – well documented in knee OA patients – were a contributing factor to the pain: it was hypothesised that the Alexander Technique lessons might reduce the amount of co-contraction (stiffness), based on evidence from earlier studies by Tim Cacciatore and others into the effects of the Alexander Technique.  Accordingly, the study investigated changes in biomechanical behaviour following the programme of lessons.  For this part of the research both the Alexander intervention group and the healthy subjects were studied walking along a test walkway. Instruments for data collection included :

  • electromyographic [EMG] sensors attached to hamstrings (flexors) and quadriceps (extensors)
  • 10 motion-capture cameras
  • two force and 3D motion plates embedded in the test walkway.  

Participants were trained to walk at a constant speed (+ or = 10%) in order to reduce confounding variables in the EMG readings and markers were fixed at anatomical landmarks on the legs to facilitate the motion capture by the cameras.

The range of instrumentation used and the extensive processing and analysis of the recorded data which the study describes illustrate the challenges involved in understanding the biomechanics of human gait, around which many uncertainties still exist – as can easily be established by reference to the sources quoted by the authors of this research.[9] Their study focused on the results from two parts of the gait cycle which previous research had indicated could be expected to be of most relevance:  a ‘pre-contact phase’ immediately before the foot hit the floor (-5% to 0% of the cycle), and an ‘early stance phase’ (15% to 25% of the cycle).


When considering the figures for co-contraction presented in the study it is important to be aware of which of the four possible measures is being reported, as they are always treated separately:

  • medial (inner side of the leg) co-contraction in the pre-contact phase
  • lateral (outer side of the leg) co-contraction in the pre-contact phase
  • medial co-contraction in the early stance phase
  • lateral co-contraction in the early stance phase

As expected from previous studies, the average EMG results in all cases, before and after, clearly showed more muscle activity in the osteoarthritis subjects than the healthy subjects. But there were observable reductions in average co-contractions across most measures  immediately after the Alexander Technique intervention[6] even if they also remained well above the levels for the healthy subjects.

Amongst the most important findings of the research was a significant but moderate correlation between medial co-contraction in the pre-contact phase of the gait cycle and the WOMAC pain scores. There was also evidence of a correlation between lateral co-contraction and pain scores in this phase, but this did not reach the generally accepted level of statistical significance of ‘p<0.05’.

Leaving aside correlations, the average reduction in medial co-contraction was 13% in this phase, although there was no reported average reduction in the amount of lateral co-contraction. In the early stance phase, there were reductions in both medial and lateral co-contraction, with the reductions in medial co-contraction achieving statistical significance.  , The research report does not report the relationship between changes in medial and lateral co-contraction in individual subjects, making it difficult to assess how this might have affected, for example, rotation at the knee.

The individual results shown in the scatter diagrams that plot changes in co-contraction against pain scores (Fig. 4 of the study[7])do not provide as consistent a picture of the biomechanical effects of the Alexander Technique lessons as one might have expected if one started from the perspective that lessons would normally lead to ‘release’, i.e. lesser tension, thus lesser co-contraction.  As noted above, there was no average reduction in lateral co-contraction in the pre-contact phase; and  as regards medial co-contraction, the 13% reduction was not distributed consistently across the participants: the scattergram that plots co-contraction in the pre-contact phase against the WOMAC pain scores, yields 7 subjects showing increased co-contraction and another 5 showing levels that are unchanged or only slightly reduced.  The reported average reduction in medial co-contraction is mainly due to the 7 subjects with noticeably larger reductions than the other 12[8]  

Of the 13 subjects whose WOMAC pain scores reduced by more than 5, 4 demonstrated increased medial co-contraction and another 3 demonstrated little change. This does not, of course,  invalidate the study's findings that medial co-contraction in the pre-contact phase was, on average, reduced, nor that there was a significant correlation between co-contraction and pain reduction, but shows it in a different light and underlines the need for caution around assuming cause-effect relationships.

Kinematics and kinetics

The outcomes reported from the kinematic analysis were sagittal (as seen from the side) knee angle (measured in degrees), sagittal plane knee moment and frontal plane knee moment (both measured in Newton metres per bodyweight Kilogram: these are measures of torque).  They are only reported in the stance phase (when the leg was loaded), not the pre-contact phase, and the changes are plotted as curves over the duration of the phase. Interpreting these curves from the perspective of the Alexander teacher who wants to understand nuances in movement is beyond the competence of the author of this news story.  Surprisingly, however, no changes are apparent between the ‘before’ and ‘after’ curves for the osteoarthritis subjects; they are almost eerily superimposed on one another, although they are easily distinguishable from those for the normal subjects. This is another counter-intuitive result from the Alexander perspective, where one might have expected a discernible change in the profile of the movement after a course of lessons. This may be because the results were too subtle to register or simply because the crucial changes occurred in aspects that the instruments were not recording, such as in the head-neck-back relationship or rotation of the leg in the transverse (horizontal) plane.


A further test involved measuring maximum muscle strength with a dynamometer: this was used in part to normalise the results from the EMG data produced in the walking trials.  As expected, the osteoarthritis subjects were significantly weaker than the healthy subjects.  In this case, the authors reported no difference between the before-and-after measures in strength for the osteoarthritis subjects.  They considered this relevant – given the reductions in WOMAC scores – for interventions that focus on strength-building, commenting:
This finding challenges current clinical management models of knee OA which focus primarily on muscle strengthening.

This would accord with the Alexander Technique view that co-ordination is more important than strength.

Anticipation of pain

Yet a further measure involved electroencephalogram [EEG] recordings of brain activity aimed at detecting whether the Alexander Technique intervention had any impact on anticipation of pain, an outcome which had been reported with other interventions such as mindfulness training and placebo. The researchers had identified other studies that had found that anticipation of pain could affect the perception of pain at the point when the pain actually occurred. As Alexander Technique lessons involve learning a range of cognitive strategies in the planning and execution of movement, the possibility existed that, simply by changing aspects connected with anticipation of pain, the perception of pain would change: this part of the study aimed to explore whether this might be the case.  

Measuring the degree of anticipation  involved recording

  • the participants’ EEG responses to actual pain induced by laser stimulators on the forearm,
  • the EEG responses to auditory warnings that the pain was about to occur (thus indicative of the level of anticipation of pain). 

The degree of stimulation was adjusted for each individual participant until they reported a pain rating of 7 on a 0-10 scale, indicating ‘clearly painful but easily tolerable’: this was their pre-intervention baseline.  After the Alexander Technique intervention, the subjects were retested with an identical pattern of auditory warnings and laser stimulation to see if there was evidence that their pain anticipation had changed. As it turned out, no statistically significant changes emerged from this part of the study.  The authors note that the absence of significant results suggests that any change in pain reporting following the Alexander Technique lessons was not down to purely cognitive changes around pain, and was thus more likely be related to the biomechanical changes.


In conclusion, this is a welcome addition to the Alexander Technique research literature. The main author, Dr Steve Preece, Research Fellow and Deputy Director of the University of Salford Centre for Health Science Research, was keen to promote the clinical significance:

Although Alexander Technique lessons have been used successfully to help people with a range of problems, this is the first time anyone has studied their effectiveness in helping people with osteoarthritis - a condition which affects nearly nine million people in the UK and which will become much more prevalent as we have an ageing population.

These sessions had a marked impact on the patients we looked at, even 15 months later, and this research could point the way towards a completely new approach for helping people with osteoarthritis.

An obvious practical implication would be that a full-scale Randomised Controlled Trial should be conducted. In fact Steve Preece, the main author, had organised a multi-site team of co-investigators and submitted a £1.2M proposal for a follow-up study, which is understood to have only narrowly failed to recived funding because of the competitive pressure for the money available, not through any deficiencies in the research proposal; it is understood that further attempts are likely to be made.[10

As for the biomechanical aspects, the significant correlation between medial co-contraction  in the pre-contact phase of the gait cycle and pain scores is an important result. The study lends weight to the view that reducing co-contraction would, more likely than not, produce benefits for patients in terms of reduced pain.[11] That said, there is not a consistent pattern whereby reductions in WOMAC pain scores were always associated with reductions in co-contraction, suggesting that the relationship between pain and co-contraction is far from straightforward. Any easy assumptions that Alexander teachers might be tempted to make about  probable uniform biomechanical effects of lessons would not be confirmed by this research. It is probable that the lessons produced bio-mechanical changes that the study did not set out to capture; it may also be that different participants responded to their lessons in different ways, possibly connected to their individual clinical condition. Whatever the explanation, the positive clinical outcomes strongly invite not only an RCT but also further in-depth bio-mechanical research to illuminate what the changes might be.


[1] Stephen J. Preece, Richard K Jones, Christopher A. Brown2, Timothy W. Cacciatore and Anthony K P. Jones, 'Reductions in co-contraction following neuromuscular re-education in people with knee osteoarthritis', BMC Musculoskeletal Disorders, 17/37, accessible via our Library from here.

[2] This comes to 24 in total, so presumably one of the figures has been misreported

[3] WOMAC was developed specifically for people with knee and hip osteoarthritis. It includes three subs-scales for pain, stiffness and function.  According to the American College of Rheumatology the WOMAC
‘is one of the most widely utilized self-report measures of lower extremity symptoms and function. It has been studied over a period of almost 30 years in many different contexts and patient populations, and there are abundant data regarding its utility and measurement properties.’ (<> accessed 8 September 2016).

[4] The study authors state that the full WOMAC inventory included 20 items whilst the WOMAC website and other literature refer to 24 items.

[5] Use of analgesics at 15 months was not recorded. The study does not report whether any of the subjects took further Alexander lessons after the trial intervention was completed.

[6] These tests were also performed immediately after the intervention but not at 15 months.

[7] Note that the titles (‘a) Medial’, ‘b) Lateral’) appearing above the two scatter diagrams are reversed in the legend to the figure (‘a) lateral and b) medial’).  Based on the discussion in the main text of the article, the former is correct.

[8] The report notes that ‘due to problems with the instrumentation, it was only possible to obtain EMG for 19 of the 21 OA participants.’

[9] See for example Richard Baker, Measuring Walking: A Handbook of Clinical Gait Analysis, (London: Mac Keith Press, 2013).

10 Personal communication from Tim Cacciatore.

[11] This is an important outcome, given that some researchers have actively challenged the desirability of reducing co-contraction: see Martijn Steultjens et al. ‘The pros and cons of muscle co-contraction in osteoarthritis of the knee: comment on the article by Lewek et al.’ and Lewek’s response, Arthritis & Rheumatology, 54/4 (April, 2006), pp.1354-5; <> accessed 30 October, 2016 [38.5Kb PDF file].

News Category: 
Latest Research


Knowledge domain: 
AT history, people, institutions: