The role of structured physiotherapy in treating patients with atraumatic shoulder instability: Medium term results from a case series

Abstract

Background

Consensus favours conservative treatment for atraumatic shoulder instability, but literature is scarce on the topic. We therefore prospectively assessed the results of structured physiotherapy for these patients.

Methods

Patient reported outcomes were recorded prior to physiotherapy and on discharge. Notes review identified patients re-referred for the same condition.

Results

N = 85. Review range was 12–72 months post-treatment. Median Oxford Shoulder Instability Score (OSIS) improved from 21 (range: 2–47) to 39 (11–47). Median Western Ontario Shoulder Instability Index (WOSI) improved from 1117 (range: 306–2028) to 485 (0–1569). Patients with posterior instability demonstrated better results compared with other groups (OSIS change, p = 0.025; WOSI change, p = 0.060). Quicker referral to physiotherapy gave improved outcomes (OSIS change, p = 0.004, r_s = −0.4; WOSI change, p = 0.047, r_s = 0.24). Twenty-one patients (24.7%) were re-referred, seven of them for repeat physiotherapy and 14 of them for surgery. Previous surgery significantly affected the possibility of a further referral (p < 0.001), and initial diagnosis was significantly correlated with further surgery (p = 0.032).

Discussion

Early referral to physiotherapy may produce better results. Patients with posterior instability responded better to physiotherapy. Previous surgery increased the risk of re-referral. Re-referred patients with posterior instability tended to be managed with further physiotherapy.

Introduction

The glenohumeral joint, due to its great mobility, is prone to becoming unstable and consequently is the most frequently dislocated joint in the body.¹ The stability of this joint during movement relies mostly on its dynamic stabilisers and neuromuscular system.² Therefore, it is often difficult to determine the boundaries between normal translation and pathological instability, which can be subtle or gross. In general, symptomatic laxity of the glenohumeral joint is defined as shoulder instability.³ Symptoms of that abnormal motion mostly include pain, a feeling of insecurity, and dislocation.⁴

Shoulder instability can be classified as atraumatic or traumatic.^5,6 Criteria for atraumatic instability are lack of traumatic incidents, absence of pain – or minor pain – during an instability incident and spontaneous relocation. However, there is no clear definition and classification of the intra-articular pathology in this type of shoulder instability. It has been advocated that a higher degree of laxity exists in atraumatic unstable shoulders, when compared with post-traumatic ones. Other causes for atraumatic unstable shoulders could be weakness of the dynamic stabilisers, neuromuscular disorders, congenital osseous dysplasia, a hypoplastic labrum, or a non-functional thinned capsule due to disturbances in the production of collagen.^7,8

Shoulder instability can have a serious impact on patients’ daily activities. Individuals with an unstable shoulder may have a reduced ability to function physically, socially and emotionally.⁹ Subsequently, the decrease of life quality ranks in severity with other major medical diseases such as hypertension, congestive cardiac failure, acute myocardial infarction, diabetes mellitus and clinical depression.⁹

Physiotherapy has been supported as the first treatment choice for atraumatic shoulder instability.^10,11 Such conservative management includes strengthening of the stabilising rotator cuff muscles and improving humeroscapular coordination with an end goal of a gradual return to normal working and sports activities.¹² However, it should be noted that in one of the few studies to analyse physiotherapy for unstable shoulders, Burkhead and Rockwood found that up to 20% of patients may still need surgery after failed conservative treatment.¹³

We searched the websites of online trial registries, ‘www.clinicaltrials.gov’, ‘http://www.isrctn.com’, ‘onlinelibrary.wiley.com/cochranelibrary’, as well as ‘www.ncbi.nlm.nih.gov/pubmed’. To the best of our knowledge, no completed randomised clinical trials or prospective cohort studies have compared operative and non-operative treatment for the management of atraumatic shoulder instability.¹⁰ Observational studies have so far demonstrated positive results from either surgical or conservative management for atraumatic shoulder instability.^13–17 However, articles showing good outcomes from specialist physiotherapy treatment for atraumatic shoulder instability are limited.^13,16,17

Aim of this study was to prospectively evaluate the results of structured specialist physiotherapy in a cohort of patients with atraumatic shoulder instability (anterior, posterior, multidirectional), using modern, robustly constructed and appropriate outcome measures.^18,19

Materials and methods

This was a single-centre, prospective case series study. It began in 2009 and stopped enrolling further patients in 2014. During this period 85 patients with atraumatic shoulder instability were included and treated by a single therapist (first author) in the institution’s physiotherapy clinic. Demographic data are shown in Table 1. Verbal consent from patients was obtained before initiation of treatment. Ethical board approval was not needed for this study as this was an evaluation of services already being applied as a standard of care for this type of patients.

Table 1.

Demographics of patients included in the study.

Type of instability	No of patients	Gender Male/Female	Agea	Previous surgery Yes/No	Previous physio Yes/No	Weeks of historya	Duration of treatment (weeks)a
Anterior	30	10/20	22 (10–46)	8/22	20/10	212 (8–852)	17 (4–60)
Multidirectional	23	7/16	20 (14–45)	4/19	14/9	131 (6–522)	32 (7–72)
Posterior	32	17/15	17 (9–31)	3/29	16/16	106 (4–887)	13 (1–129)
Total	85	34/51	20 (9–46)	15/70	50/35	157 (4–887)	17 (1–129)

^aNumbers are medians with range in parentheses.

This study was designed as to obtain patient reported outcomes prior to treatment initiation and analyse changes as recorded at the day of each patients’ discharge from physiotherapy. Further follow-up was needed after discharge in order to check if patients were re-referred for the same condition (minimum: 12 months, maximum: 72 months). The type of treatment (physiotherapy or surgery) that patients received after re-referral was recorded.

For analysis purposes, instability was defined as anterior, posterior, or multidirectional if it involved two or more planes (anterior and/or posterior instability accompanied by inferior subluxation).²⁰ Diagnosis was usually made by the referring consultant shoulder surgeon and confirmed in all cases by the specialist physiotherapist. Inclusion criteria for this study were patients referred to the physiotherapy clinic and diagnosed with atraumatic shoulder instability. Patients with diagnosed psychiatric disorders, traumatic instability, prior injury or fracture to the affected shoulder and ipsilateral clavicle or proximal humerus, congenital deformity or severe neuropathic disorders were excluded from the study.

The Oxford Shoulder Instability Score (OSIS) and the Western Ontario Shoulder Instability Index (WOSI) were completed during the initial and final (discharge) physiotherapy sessions. The OSIS is a short, 12-item, condition-specific, patient reported outcome (PROM) measure developed and validated for measuring surgical and non-surgical therapeutic outcomes of patients presenting with unidirectional or multidirectional instability of the shoulder. It scores from 0 (worst score) up to 48 points (best score).²¹ The WOSI questionnaire consists of 21 items, each scored on a 100-mm Visual Analogue Scale. Each item falls into one of the domains of Physical function, Sports/Recreation/Work, Lifestyle and Emotional well-being. Each question is scored between 0 and 100 points and the summation of all questions results in a final WOSI score. The final score ranges from 0 (best possible score – the patient is experiencing no decrease in shoulder related quality of life) to 2100 (worst score – signifies extreme distress in shoulder-related quality of life).²²

Structured physiotherapy treatment approach

Initially, the direction of instability, frequency, ease of subluxation and ease of reduction were determined. Where necessary, this was determined by examination (often, the patients would willingly demonstrate themselves), with the apprehension test, posterior jerk test and the load and shift test, as required. Many patients were pain free and could easily self-reduce subluxations, but great care was taken examining patients who might need a manipulation under anaesthesia (MUA) to reduce a subluxation. Indeed, several patients presented from the emergency department (ED) with subluxed shoulders after a failure to stay reduced after MUA. They had often undergone several attempted reductions, and had subluxed again before getting out of theatre. One presented with simultaneous bilateral subluxations, one anteroinferior (a partial luxatio erecta, in over 90^o of abduction) and the other posterior.

The treatment approach is presented in Table 2 and can be summarised as educate, mobilise, activate, control and perform. We found that most patients had a stressful journey to our services, had often been treated (unsuccessfully) as traumatic instability in ED, and had been given conflicting advice by relatives, friends, the internet and healthcare professionals, most of whom would be unfamiliar with this rare condition.

Table 2.

Structured physiotherapy approach.

Stage	Goal	Activity	Load suggestion
Educate	Patient ± family to understand management of atraumatic instability	Discussion	NA
Mobilise	Unrestricted active movement of thoraco-scapulo-humeral system	Topical heat, massage, but predominantly self-stretch	As tolerated by patient
Activate	Able to produce force in thoraco-scapulo-humeral system without subluxation/pain	Predominantly isometric exercises in safe zones	Aiming for 10 reps, 10 second holds, 4 times daily. Adapt to patient convenience and response
Control	Able to move arm freely, with moderate load and without subluxation/pain	Predominantly dynamic exercises	Aiming for 10 reps, 4 times daily. Adapt to patient convenience and response
Perform	Full function without subluxation/pain	Bespoke exercises	Build up to patient requirements

Education of the condition, its prognosis and management is vital as in most cases self-management is the optimum treatment. Rather than having ‘something wrong’ with them, we encourage patients to regard themselves as ‘high performance’, like a thoroughbred racehorse or a Ferrari – capable of great things, but needing some skilled handling to bring the best out.

Mobilising aims to release any comparative soft tissue stiffness, which may restrict thoraco-scapulo-humeral movements, and would include reduction of chronic subluxations (our longest had been in a sling, out of joint, for six months prior to treatment).

Activate involves encouraging the musculature to contract in a similar manner to the ‘normal’ side, whilst staying in joint.

Control is then moving throughout normal ‘safe zone’ range whilst controlling thoraco-scapulo-humeral alignment, i.e. pain-free and in joint.

Performance is bespoke, and involves building the range, load, speed, frequency and duration as per individual needs.²³ This has involved mimicking martial arts, kayaking, rugby tackles, javelin, swimming, dancing, breast feeding, free-fall parachuting and hugging a parent amongst other activities.

Statistical analysis

Statistical analysis of previous relevant literature was used in order to identify the number of patients needed for this study. Burkhead and Rockwood found that 20% of patients treated non-operatively for atraumatic shoulder instability finally required operative treatment.¹³ Takwale et al. similarly found that 12/58 shoulders (20.6%) relapsed and required further retraining.¹⁶ Sample calculations with a priori power analysis with G* Power²⁴ used OSIS score as a primary outcome measure. We hypothesised that 20% of patients final scores may be low (0–19 OSIS score) and that these patients will need either retraining or surgical referral after discharge from clinic. With a 95% probability of all other patients having better scores, α error = 0.05, Power = 0.8, the estimated sample size of this study was 71 patients. For an expected 20% loss of follow-up, the needed final size of this study rose to 85 patients.

Normal distribution was analysed by using the Shapiro–Wilk test. As normality assumptions failed, quantitative parameters were studied with the Mann–Whitney U test; for subgroup analysis, the Kruskal–Wallis H test was used in order to identify any differences between groups. The Chi-square test was used for analysis of qualitative variables. The Spearman test was used in order to identify significant correlations between quantitative non-parametric variables. The Wilcoxon signed rank test was done in order to assess correlations between initial and final scores. Statistical differences were considered significant for p < 0.05.

Results

The median age of patients was 20 (minimum 9–maximum 46). Sixteen patients had previously undergone surgery for the same condition and were included in the study. We were unable to retrieve the operative details of these patients. Anterior shoulder instability was identified in 30 patients, posterior instability in 32 and multidirectional instability in 23 patients. Median treatment period was 17 weeks (range 4–129 weeks).

All patients were able to provide the outcome scores at both time points of this study. Results of the pre- and post-treatment OSIS and WOSI scores are shown in Figure 1(a) and (b). Median OSIS improved 18 points (37.5%) from 21 (range: 2–47) to 39 (range 11–47). Median WOSI improved 632 points (30.10%) from 1117 (range: 306–2028) to 485 (0–1569). All types of shoulder instability were significantly improved at final follow-up (p < 0.001).

Figure 1.

(a) Box-plot chart comparing the first and final OSIS scores between different subgroups of patients with atraumatic shoulder instability. (b) Box-plot chart comparing the first and final WOSI scores between different subgroups of patients with atraumatic shoulder instability.

We calculated the differences between the first and final OSIS and WOSI scores, in order to identify any correlations between the change of these scores and other variables. Patients with posterior instability demonstrated better change in outcome scores when compared with other groups (OSIS change, p = 0.025; WOSI change, p = 0.060). Previous surgery did not significantly affect the final outcome scores. Also, age, duration of treatment and previous physiotherapy did not significantly affect the final outcome scores. The final OSIS and WOSI score and their changes (final minus initial) were as expected strongly negatively correlated (p < 0.001).

Comparing weeks of history (time span of symptoms) with the OSIS and WOSI changes, showed that patients appeared to do better if they were referred sooner for physiotherapy (OSIS change, p = 0.004, r_s = −0.4; WOSI change, p = 0.047, r_s = 0.24). However, patients’ length of symptomatic history did not significantly affect the rate of re-referral.

Twelve months after the last study’s patient discharge from the clinic, we searched for any re-referrals for this group of patients to either a physiotherapy or orthopaedic shoulder clinic on our institution’s catchment area. A total of 21 patients (24.7%) were further referred, seven of them for repeat physiotherapy and 14 of them for surgery (Table 3). This additional need for treatment was not significantly correlated with any outcome score or their difference (calculated from final minus first score), nor with age, gender or initial diagnosis.

Table 3.

Patients being re-referred after discharge from the physiotherapy clinic; age, previous surgery to the affected shoulder, type of instability, the number of months it took for them to be re-referred and final treatment received are shown.

Patient No.	Age	Previous surgery	Instability type	Months after discharge	Final treatment received
1	24	No	Anterior	24	Physiotherapy
2	34	Yes	Anterior	12	Surgery
3	29	No	Anterior	3	Surgery
4	31	Yes	Anterior	3	Surgery
5	32	Yes	Anterior	3	Surgery
6	24	Yes	Anterior	3	Surgery
7	28	No	Anterior	6	Surgery
8	14	Yes	Anterior	1	Surgery
9	18	No	Multidirectional	96	Physiotherapy
10	22	Yes	Multidirectional	1	Surgery
11	15	No	Multidirectional	1	Surgery
12	21	Yes	Multidirectional	4	Surgery
13	18	No	Multidirectional	4	Surgery
14	33	Yes	Multidirectional	1	Surgery
15	14	No	Posterior	28	Physiotherapy
16	16	Yes	Posterior	1	Surgery
17	23	No	Posterior	1	Surgery
18	18	Yes	Posterior	15	Physiotherapy
19	14	No	Posterior	3	Physiotherapy
20	15	No	Posterior	30	Physiotherapy
21	21	No	Posterior	25	Physiotherapy

However, it was found that previous surgery for the same condition before entering this study’s physiotherapy approach, significantly affected the possibility of a further referral (p < 0.001). From the 16 patients who underwent an operation before entering this study, 10 of them were re-referred and nine out of these 10 underwent further surgery, usually capsular plication. The latter was also statistically significant (p = 0.031).

We also analysed the type of treatment that patients had after re-referral for possible correlations with any outcome score, or another variable. We found that initial diagnosis was significantly correlated with the type of final treatment (p = 0.032); 7/8 patients with anterior instability, 5/6 patients with multidirectional and 2/7 with posterior instability underwent surgery after re-referral (Figure 2). As shown in Table 3, almost half of the patients receiving surgery (6/14) were referred back within a month of referral from orthopaedic clinic, 12/14 within four months, and all of them were referred back within a year. All of these referrals (except from one which took 12 months to be re-referred and operated) were made by our own senior physiotherapist. For these cases, it was decided that the patients clinically appeared to have such structural instability of the shoulder that they would benefit from operative treatment. All other re-referrals to either surgery or physiotherapy were done due to the patients having a major incident of shoulder instability/subluxation that caused them pain and difficulty to carry on successfully with their daily activities.

Figure 2.

Bar chart demonstrating the number of patients undergoing further physiotherapy or surgery after being re-referred, grouped by their initial diagnosis.

Discussion

Non-operative treatment in terms of physiotherapy has long been advocated for atraumatic shoulder instability.²⁵ The condition is rare,²⁶ and this may be one of the largest reported cohorts with modern outcomes assessment¹⁸ to date. Burkhead and Rockwood studied the results of specific physiotherapy exercises on 140 patients, 74 with a traumatic and 66 with atraumatic onset.¹³ They demonstrated that better results were achieved from patients with atraumatic shoulder instability (53/66 patients, 80%). Takwale et al. diagnosed 58 shoulders with involuntary positional instability.¹⁶ The results were positive in 46 shoulders for patients using a specific exercise protocol; 12 shoulders relapsed and required further episodes of retraining. Bateman et al.¹⁷ published the outcomes of 18 patients with a mean follow-up of 4.5 months. There was only one patient with posterior instability. The article favoured the use of a structured programme for patients with atraumatic shoulder instability. The mean WOSI improved by 36.76% and the mean OSIS by 16.67 points (34.73%).¹⁷ Watson et al.²⁷ and Warby et al.²⁸ reported 17 points (35.42%) OSIS and 793 points (37.76%) WOSI changes at 4.6 months (n = 43), and 34.9% WOSI change at 24 weeks (n = 41) respectively for a structured MDI rehabilitation programme. Our study has comparable results, with a much larger, heterogeneous cohort, and longer follow-up, and adds to the evidence that positive outcomes can be achieved in routine practice for patients with atraumatic shoulder instability, by using structured physiotherapy. This may include work on scapular control, however, we have found that this is not required for most patients, if other issues are adequately addressed. Although we do less and less scapular work, in our clinical experience there are a small group of patients who struggle to improve without tackling the scapula. These are identified by elimination of symptoms when the closed mechanism of the shoulder girdle spatial mechanism²⁹ is restored by facilitation. For this reason, we would suggest that the term scapular dyskinesia is used only when the movements are symptomatic, similar to the distinction between laxity and instability in the glenohumeral joint.

However, patients being referred to physiotherapists with atraumatic shoulder instability are not guaranteed to lack any intraarticular lesions. Werner et al. evaluated by diagnostic arthroscopy 43 patients with atraumatic shoulder instability, not responsive to physiotherapy.⁸ All of these patients had a pathological lesion; 13/43 (30.2%) had classical Bankart lesion and 11/43 (25.6%) had complex labrum/capsule lesions. It is unknown if these lesions would exist on such spectrum if a firm diagnosis was made sooner and appropriate referral was done to a specialist physiotherapy clinic, as a quicker referral positively correlated with better outcomes. One however should note that if a patient is unresponsive to conservative treatment, then it is possible that he has a structural lesion, which could be repaired by a surgical intervention. We found that patients having undergone surgery prior to being included in this study were more likely to be re-referred and subsequently undergo further surgery.

There is a distinction between habitual, voluntary and involuntary instability and there is a potential confusion regarding these terms as they all have been used in the past without a sufficiently clear definition.¹⁶ Rowe et al. used the term voluntary for all of his patients who could dislocate one or both shoulders voluntarily by doing specific movements.³⁰ Huber and Gerber analysed the outcomes of 25 children (36 shoulders) and used the term ‘voluntary’ with the word habitual next to it to describe the way these children were able to dislocate their shoulder. This group of patients had no psychiatric disorders. They concluded that no restriction of activity and no physiotherapy was better than any form of operative treatment in terms of outcomes and complications.³¹ Takwale et al. studied a group of patients only with involuntary shoulder instability and gave the term ‘involuntary positional instability’.¹⁶ They highlighted that in 31/50 patients, the condition had not been recognised before referral and concluded that there should be increased awareness of the condition. In our study, we excluded patients suffering from known psychiatric disorders but included patients being able to dislocate their shoulder deliberately themselves. The correlation between symptomatic weeks before referral and final outcome scores indicates that the quicker a correct referral and diagnosis are made, the better the final outcomes from these patients maybe.

Patients with significant atraumatic laxity of the posterior capsule may complain of posterior instability. Shoulder elevation, horizontal adduction and excessive internal rotation place excessive stress on the posterior capsule in these positions.¹² In Burkhead and Rockwood’s study patients with posterior instability responded better than those with anterior instability.¹³ The authors hypothesised that symptoms in that subgroup of patients have originated from ligamentous and capsular laxity and not osseous architectural changes. Final outcomes from our study demonstrate similarly that patients with posterior instability may respond better than others with non-operative treatment and also have less chances of undergoing surgery if they get re-referred for the same condition.

This study has several limitations. It does not have a control/sham group or a comparison with a surgical group and we do not know how these patients would respond to an alternative treatment. It is a case series study and as such, since it is not randomised or blind, it could be positively biased. However, the outcome scores are patient reported and the physiotherapist could not influence the results. All 85 patients filled the initial and final outcome scores, and the study is well powered for its primary endpoints which is the measurement of OSIS and WOSI scores. It should be noted however that for results that have been produced from subgroup analysis, such as outcomes related to a specific diagnosis (anterior/posterior/ multidirectional instability, previous surgery), the study may be underpowered. Therefore, these results should be interpreted with caution. Regarding surgery, we were unable to identify the specific type of operation that these patients had done before. However, it should be noted that due to the small number of this subgroup (16 patients) further sub-analysis would be even more underpowered; the possible results would be inadequate for statistical interpretation. All patients’ physiotherapy exercises were supervised by a single specialist physiotherapist on a single centre. There was no deviation from treatment resulting from different physiotherapists. Since criteria for re-referral was based on our institution, it is possible that dissatisfied patients, or patients moving out of our locality were not accounted for. The authors believe that currently there is no clear alternative to physical therapy, regarding conservative treatment, and an operation may only be indicated when a pathological structure can be surgically treated. A control group would probably consist of patients being allowed to do anything and not receiving any form of therapy, however the long symptomatic history of some of our patients suggest that this may not produce improvements. Physiotherapy has consistently produced positive results for the conditions mentioned in this article. Therefore, a future similar study could probably focus on analysing outcomes of different physiotherapy regimes and the patient and analysis investigator be blinded to the regime used.

Conclusion

Non-operative treatment in terms of structured physiotherapy may yield successful outcomes for atraumatic shoulder instability, with patients achieving good/ excellent results. Early referral to physiotherapy may produce better final results. Patients with atraumatic posterior instability seem to respond better to physiotherapy when compared to patients having anterior or multidirectional instability. Previous surgery for the same condition increased the risk of re-referral and subsequently a further operation. Most patients being re-referred in our study for the same condition either had a structural disorder that needed operative treatment or suffered another major instability/subluxation incident that warranted further physiotherapy and rehabilitation.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Ethical approval

No ethics committee approval was needed for this study.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.