Shoulder instability or more specifically glenohumeral instability is a common and well described problem encountered in sports injuries. However its treatment and the classification of the spectrum of injuries to guide that treatment has been problematic in the sense that traditional systems of classification have not been comprehensive and inclusive of all the pathologies that contribute to instability. This is particularly true of muscle patterning disorders. This article reviews how the classification systems have developed to include this often overlooked cause of instability and underlines current thinking on the treatment of muscle patterning disorder.


There are three key components which confer stability on the shoulder joint. First, there is the conformity of the humeral head on the smaller glenoid cavity which is further deepened by its labrum. Second there is the negative pressure suction effect within the capsule of the joint which relies on capsule integrity and third there is the compression of the head onto the glenoid cavity by various muscles around the joint including the rotator cuff muscles. All three combined provide stability but rather interestingly absence of some or one of these factors does not invariably lead to instability. So in order for instability to occur there are several possibilities. The structural components may be deficient, and this can be due to either a traumatic event or an atraumatic phenomenon. The muscular activity and coordination may be impaired or the third possibility is that the joint must be subjected to a force beyond the capacity of the system to absorb whilst retaining its integrity.

There are however another two additional factors to be considered. A feature not often examined is the effect of time. The classic young patient presenting with instability often cites a significant traumatic event for the first dislocation but subsequent dislocations present with increasingly minimal trauma. This suggests that the type and degree of severity of instability can change with time and this also needs to be considered in order to ensure all the pathologies are correctly identified and treated at the time of the intervention. It is not uncommon to see shoulder instability being treated by an operation designed to counter the cause of instability from the initial presentation when often the pathology has become more complex with time. The second factor is the presence of hyperlaxity. This must be distinguished from instability as hyperlaxity is a structural abnormality in the composition of the tissues allowing different tensile and contractile forces to act on joints. The usual cause for hyperlaxity is a change in the composition (abnormal proportions of collagen commonly) of the soft tissues. This can create confusion in the clinical picture by contributing to the overall effect of instability.

The Matsen Classification

Medical students, junior doctors and junior physiotherapists will be perhaps most familiar with this classification. This classification divided patients into two groups - TUBS and AMBRI.

  • T Traumatic
  • U Unilateral
  • B Bankart lesion
  • S Surgery

This classification divided patients into two groups, one having a traumatic structural defect and another having atraumatic structural defect. The classic TUBS patient had a traumatic injury, which affected one shoulder and one could identify the Bankart lesion (tear of the labrum which normally deepens the glenoid cavity) and the Hill-Sachs lesion (defect of the posterior part of the humeral head which impacts the acromion during an anterior dislocation). These patients were deemed to require surgery to repair the Bankart's lesion and then all would be well. For patients presenting with anterior glenohumeral instability this is more or less true but the classification made no account of the direction of the instability.

  • A Atraumatic
  • M Multidirectional
  • B Bilateral
  • R Rehabilitation
  • I Inferior Capsular Shift

The AMBRI patient had some form of laxity of the capsule which was never precisely defined, but this required rehabilitation in the form of physiotherapy (again never precisely defined as to which type of therapy) with those not responding after a while going to have an inferior capsular shift to tighten up the capsular laxity.

This then has been the mainstay of treatment for shoulder instability for the last three decades. The shortcomings of this system is that it doesn't include hyperlaxity, you cannot combines the features of traumatic and atraumatic instability coexisting within the same patient, and neither does it allow for the change in pathology with time. Although it broadly divides shoulder instability in the two major divisions, in fact they probably represent two extremes of the clinical picture.

The Gerber Classification

In order to address some of these issues, Gerber proposed a classification to include hyperlaxity as an additional factor. He devised six types of instability.

  1. Locked / Chronic Dislocation
  2. Unidirectional Instability - without hyperlaxity
  3. Unidirectional Instability with hyperlaxity
  4. Multidirectional Instability with hyperlaxity
  5. Multidirectional Instability without hyperlaxity
  6. Voluntary instability

Roughly speaking type II instability was equivalent to the TUBS group in the Matsen classification and type IV corresponded to AMBRI group provided there was no voluntary control of the dislocation. Type III patients differed from type IV/AMBRI patients as they had recurrent painful dislocations associated with structural defects of the ligamentous complex but did not respond well to surgery. Type V patients had two or more structural defects turning them from a type II/TUBS into a type V and required surgery, however it was noted that this would be difficult in view of the plural pathology, but it was at least acknowledged that the possibility of multiple pathology exists. Type VI patients were the type IV/AMBRI patients whom had voluntary control.

This concept of voluntary control of dislocation is interesting in several respects. First, if you leave these patients alone it does not result in the development of degenerative change within the joint and second, some type IV patients could actually gain control over their dislocations and turn themselves into type VI patients further clouding the picture. This therefore although not formally classifying the effect of time on pathology, did however allow for it to be described. The concept of voluntary control was proposed by Rowe some 36 years ago. At the time, it was thought that there may be either a psychological or even psychiatric element to the condition in terms of whether it was a case of wilful behaviour for secondary gain or whether it was a case of a habitual phenomenon implying a learned response triggered by a certain situations. Both these terms are now out of favour due to a lack of evidence available to link these dislocation patterns with any established psychiatric illness or prevalence of psychiatric disease. Considering involuntary control, it is postulated that the cortical representation of normality for the shoulder changes when the shoulder when it is not correctly located and when it is correctly located the brain perceives it as being in an abnormal position. This would explain why in some patients with persistent subluxation spontaneously correct under a general anaesthetic only to sublux again after they wake up. Therefore patients labelled with involuntary control may just have a faulty sequence of muscle firing.

Muscle Control

The key works here were performed at the Royal National Orthopaedic Hospital at Stanmore by Bayley initially over twenty years ago and continued by his co researchers. Between 1983 - 2003, dynamic electromyography was performed in latissimus dorsi, pectoralis major, anterior deltoid and infraspinatus in 1097 patients with glenohumeral instability. The key findings of this landmark work were that about 50% of all glenohumeral instability patients had abnormal muscle patterning. Further analysis showed posterior instability had about 85% abnormal muscle patterning associated with it and about 25% involvement with anterior instability. 100% of patients with inferior instability had muscle patterning disorder. The significance in these findings are that if you operate on a patient for instability, who has an abnormal muscle patterning disorder you are five times (anterior instability) or ten times (posterior instability) more likely to get a poor result. So these patients must be identified to prevent an unsuccessful procedure.

Of the total of 1097 patients, 323 had a pure patterning disorder and 161 had a mixed picture of structural (either traumatic or atraumatic) defect and muscle patterning disorder. This specific group needs to be identified as conventional physiotherapy with strengthening exercises doesn't seem to work well. However using biofeedback techniques, specialist physiotherapy has a much better outcome more of which is discussed later. It is the current concept that one should always deal with the muscle patterning problem first before surgery otherwise abnormal muscle forces will interfere with the success of a surgical repair by subjecting the joint to abnormal force couples.

It is also worth noting that patients who have pure muscle patterning disorder tend to be young with a trimodal distribution of onset. The three peaks are at 6yrs, 14yrs and 20yrs with a mean age of onset around 14yrs. In the under 10yr group the condition was present more commonly in girls.

Treatment of muscle patterning disorders

In Burkhead and Rockwood's paper, they looked at 140 shoulders will all variants of glenohumeral instability. Treatment consisted of graduated muscle strengthening exercises. If there was traumatic subluxation then only 16% reported a good or better result. However if there was atraumatic subluxation then this increased to about 80%. This ties with common practice which advocates surgery for the traumatic structural causes of instability and reserves physiotherapy treatment for atraumatic cases. Kiss assessed a programme including visual feedback, proprioceptive treatment and glenohumeral and scapulothoracic pattern correction. With a sample size of 62 patients all with atraumatic instability, he reported a 61% figure for abolition of instability.

This is the most successful report I have in the literature for abolition of instability. Takwale found 52 out of his 58 patients had an excellent result at 2yrs with his muscle retraining programme but interestingly nine patients relapsed and required retraining. This suggests that further studies looking at long term follow up of muscle retraining and feedback programmes need to be performed if there is a risk of relapse following the initial treatment. Reid performed a randomised controlled trial with 20 patients and compared isokinetic resistance exercise with an EMG biofeedback re-education programme in athletes with anterior instability. The biofeedback programme performed better in terms of shoulder function both at work and in sport. Bayley's Stanmore group showed an 88% improvement with a biofeedback programme in patients with muscle patterning instability but only 34% actually achieved a stable shoulder which is something that should be borne in mind when counselling patients.

Bayley's Triangle

Given these findings, it is clear that the previous classifications will not aid correct treatment as the whole spectrum of causes of instability is not included. Therefore a new classification was required and was pioneered at Stanmore by Bayley. This is the most current classification in use and it neatly ties in aspects of the previous classifications whilst extending them to include the plurality of pathology and the effect of time.

  • Traumatic
  • Atraumatic Structural
  • Muscle Patterning

Based on the concepts discussed above, there are three groups of causes of instability. One, two or all three can be present in a patient and indeed can change over time. A patient can be placed anywhere within the triangle to reflect the contributions of each group of causes and if the triangle is projected into a three dimensional shape, one can plot the effect of time on the relative contributions of each factor.

  1. Extrinsic Force Traumatic
  2. Intrinsic Atraumatic Structural (including hyperlaxity)
  3. Muscle Patterning

The apices are labelled I, II and III as described above. As you go from one apex to the other, the relative contributions vary. As such the lines connecting apices are further labelled with the relative contribution. From the connecting line between apices, depending on the contribution of the opposite apex, one places the patient towards that opposite apex. The final axis is that of time. One can turn the triangle into three dimensions and plot the patient at time intervals from initial presentation to demonstrate how relative contributions from each factor changes.


Ultimately, a classification system should aid achievement of the best possible outcome by guiding treatment. The Stanmore classification currently provides the most comprehensive system for both simple and complex cases of instability. It caters for patients with muscle patterning disorders both of a pure nature and in combination with structural abnormalities. Most importantly however it helps to guide treatment towards those who require specialist physiotherapy and those who require surgery.


  1. The Classification of Shoulder Instability. New Light Through Old Windows. Lewis A, Kitamura T, Bayley JIL, Current Orthopaedics 2004 18:97-108
  2. The Classification of Glenohumeral Instability. Gerber C, Nyffeler RW, Clinical Orthopaedics & Related Research 2002 400:65-76
  3. Evaluation and Classification of Shoulder Instability With Special Reference to Examination Under Anaesthesia. Cofield RH, Irving JF Clinical Orthopaedics & Related Research 1987 223:32-43
  4. Treatment of Instability of the Shoulder with an Exercise Program Burkhead WZ Jr, Rockwood CA Jr, Journal of Bone and Joint Surgery (Am) 1992 74A: 890-6
  5. An Approach to the Repair of Avulsion of the Glenohumeral Ligaments in the Management of Traumatic Anterior Glenohumeral Instability. Thomas SC, Matsen FA III, Journal of Bone and Joint Surgery (Am) 1989 71A: 506-13
  6. The Role of Dynamic Electromyography in Muscle Patterning Instability. British Elbow & Shoulder Society Annual Meeting Edinburgh 2006
  7. Non-operative Treatment of Multidirectional Instability. Kiss J, Damrel D, Mackie A, Neumann L, Wallace WA, International Orthopaedics 2001 24:354-7
  8. Involuntary Positional Instability of the Shoulder in Adolescents. Is There Any Benefit from Treatment. Takwale VJ, Calvert P, Rattue H American Journal of Sports Medicine 2000 82: 719-23
  9. Anterior Shoulder Instability in Athletes: Comparison of Isokinetic Resistance Exercise and an Electromyographic Biofeedback Re-education Program. A Pilot Program. Reid DC, Saboe LA, Chepeha JC, Physiotherapy Canada 1996 Fall: 251-6
  10. Voluntary Dislocation of the Shoulder. A Preliminary Report on a Clinical, Electromyographic and Psychiatric Study of Twenty -Six Patients. Rowe CR, Pierce DS, Clark JG, Journal of Bone and Joint Surgery (Am) 1973 55A: 445-60
  11. The Classification of Shoulder Instability. Lambert SM, The Evidence for Orthopaedic Surgery Ch 8: 99-108 TFM Publishing 2007
  12. Rehabilitation of the Shoulder. Gibson, JC, The Evidence for Orthopaedic Surgery Ch 11: 130-132 TFM Publishing 2007