Post-surgical physiotherapy in frozen shoulder: A review

Abstract

Despite its prevalence, the optimal management of frozen shoulder is unclear. A range of conservative measures are often undertaken with varying degrees of success. In cases of severe and persistent symptoms, release procedures which could include any combination of manipulation under anaesthetic, arthroscopic capsular release or hydrodilatation are frequently offered, none of which has been shown to offer superior outcome over the others. When surgical release is performed a period of rehabilitation is normally recommended but no best practice guidelines exist resulting in considerable variations in practice which may or may not directly affect patient outcome. During this narrative review, we hypothesise that these differing responses to treatment (both conservative and surgical options) are potentially the result of different causal mechanisms for frozen shoulder and may also suggest that post-release rehabilitation may need to take this into account.

Introduction

When disagreement exists over the diagnostic label of a clinical condition, there is a good chance that there will be dispute over its management. This remains the case with frozen shoulder, which despite being defined in the 19th century remains an enigmatic disorder with its pathogenesis, diagnosis, prognosis, optimal treatment strategy and even its name giving cause for debate.

The severe and often debilitating pain and stiffness experienced by patients with frozen shoulder can invade every aspect of normal life. This can grossly restrict function and have a profound effect on an individual’s ability to cope with normal daily activities. The resultant disability, combined with sleep deprivation and anxiety ¹ is such that patient’s perception of frozen shoulder as a detriment to their general health has been ranked as highly as that of those suffering with congestive heart failure, acute myocardial infarction, diabetes mellitus and depression. ² The population prevalence of frozen shoulder remains largely unknown but point prevalence studies suggest between 1% and 10% of the general population^3–5 suffer from frozen shoulder but the prevalence may rise to as much as 36% in a diabetic population.^6,7 It is more common than other musculoskeletal upper limb conditions such as De Quervain’s tenosynovitis and Tennis Elbow. ⁵ In addition to the prolonged duration of symptoms experienced, there is a considerable burden placed on primary and secondary care services. ⁸ Furthermore, as most patients affected are in their 5th and 6th decade ⁶ there are socio-economic implications due to work absenteeism^8,9 and reduced productivity, which means effective management of this condition is of great importance.

Currently, there is no agreed standard treatment for frozen shoulder. Debate remains as to whether this is a self-limiting condition, whether conservative measures are preferred over surgical intervention, whether one form of surgery is more beneficial than another and indeed that frozen shoulder may not be a single entity but have various clinical phenotypes that requires stratified treatment. Patients with the most severe and prolonged symptoms often end up undergoing a form of release procedure. When they do, a period of rehabilitation is often recommended but there are no best practice guidelines and considerable variations in practice exist which may or may not directly affect patient outcome.

This review will focus on the evidence regarding the role of post-surgical physiotherapy, which as yet has received little attention in the literature on the management of frozen shoulder. With the current lack of quality evidence on which to base a more formal review, the purpose of this article is to highlight the many uncertainties surrounding frozen shoulder, to promote discussion surrounding common current clinical practices and to suggest direction for future research.

Pathophysiology

The characteristics of frozen shoulder were originally described in 1872 by Duplay who described the condition as a “scapulohumeral periarthritis”. ⁹ The more recognised “frozen shoulder” as a descriptive diagnosis was made in 1934 ¹⁰ followed by “Adhesive Capsulitis” a decade later. ¹¹ “Contracted (frozen) Shoulder” ¹² and “Frozen Shoulder Contracture Syndrome” ¹³ have been offered as the 21st century’s attempt to capture the essence of what is a painful and debilitating condition which often leads to prolonged functional restriction and disability.

“Adhesive Capsulitis” has remained a popular term despite it being a debatable description of the pathology. References to dense adhesion formation as a result of inflammation are readily found within the literature ⁹ but others suggest that no true adhesions are present. ⁶ Histological studies of affected shoulder capsule have identified fibroblast activation and proliferation together with increases in markers that would suggest the presence of inflammation and immune cell activity including mast cells, T cells, B cells and macrophages^14,15 but to suggest that inflammation is the sole, primary cause of symptoms is to perhaps oversimplify what is a complex interaction of multiple processes. Recent work^8,13,16 suggests that the frozen shoulder process involves a combination of inflammation, be it acute or chronic, and capsular fibrosis, the net result of which is contracture. Whether these processes occur in sequence, tandem or in isolation are undetermined.

Although there are strong associations between frozen shoulder and other morbidities – particularly type one and type two diabetes, thyroid disease, cerebrovascular disease, coronary artery disease, autoimmune disease and Dupuytren’s disease^6,16–18 there are limited offerings within the literature as to whether the pathology of frozen shoulder varies depending on the presence of co-morbidities. This is also true of whether the frozen shoulder is regarded as primary (insidious) or secondary (as a result of injury or trauma).

Clinical presentation and diagnosis

Currently, there are no formally recognised diagnostic criteria, ¹⁹ and what constitutes a positive clinical finding for frozen shoulder is equivocal. ¹³ Patients tend to present with insidious onset of pain which can be diffuse in nature and affect a widespread area encompassing the deltoid region, upper arm and wider shoulder girdle. A correspondence based Delphi study involving 70 experts suggested that the loss of active and passive movement, the strong presence of night pain and lying on the affected side, and sudden or unexpected movements resulting in an increase of pain were considered clinical identifiers for early stage frozen shoulder. ²⁰

Following an email survey of 211 members of the American Shoulder and Elbow Surgeon a consensus definition for frozen shoulder was developed as a “condition characterized by functional restriction of both active and passive motion for which radiographs of the glenohumeral joint are essentially unremarkable except for the possible presence of osteopenia or calcific tendonitis”. ²¹ Others have set reduction of passive range of movement of more than 30% in two out of three shoulder movements as an indicator of frozen shoulder ²² or if passive external rotation was less than 30° of the unaffected side. ²³ The recent United Kingdom Frozen shoulderTrial (UK FroST) ²⁴ required passive external rotation that was less than 50% of contra-lateral side. All the definitions described above require plain radiographs to be normal, although primary care guidelines may not always require imaging to suggest a working clinical diagnosis of frozen shoulder.^25,26

The clinical course of frozen shoulder

If a lack of consensus exists regarding the diagnostic criteria of a frozen shoulder, there is total ambiguity regarding the process and duration of this troublesome condition. The familiar three stage model originally proposed by Reeves ²⁷ suggests pain, stiffness and recovery phases which occur consecutively. This was based on a prospective secondary care study of 49 patients of whom 41 were followed up for more than five years.

The three stages – freezing, frozen and thawing have been described as both overlapping ⁶ and consecutive. ²⁸ Phase one (or the freezing stage) is reported to last anywhere from eight weeks to three months. Phase two (the frozen stage) between 16 weeks and 1 year and phase three (the thawing stage) between 16 weeks and over 3 years.^6,9 Studies providing this evidence were a mixture of retrospective comparative, prospective cohort, case series and randomised controlled trials (RCTs) with patient numbers ranging from 40 to 356 with follow-up from 1 to 10 years. Very few, if any of these studies have been conducted in primary care, where the spectrum of disease and duration of symptoms might be different. Acknowledging the inconsistencies of the three stage model Hanchard et al. ²⁹ suggested the characteristics of “pain predominant” or “stiffness predominant” can be used as an alternative to the description of the phases of frozen shoulder. However, the timespans of these alternative phases remain variable and overlap of stages persists.

Despite this lack of clarity, frozen shoulder is widely referred to as a self-limiting condition that resolves spontaneously over time.^9,18,30 Evidence as to whether this is as benign a condition as is suggested however is divided and outcomes from long-term studies report ongoing pain and stiffness over many years. ³¹

The clinical course of frozen shoulder has been explored in a recent systematic review. ¹⁹ Despite the low methodological quality of the studies reviewed, large-scale retrospective cohort studies within the analysis did identify that between 40% and 56% of patients continued to report ongoing pain and loss of range of movement for up to 20 years after the onset of their symptoms. The most frequent complaint was pain sufficient to cause night waking that occurred in up to 35% of patients and a small group (6%) rated their pain as severe.

So, although many people recover within 6–12 months, there is therefore enough substance to tentatively challenge the long and deeply held belief that the history of frozen shoulder passes through painful, stiff and recovery phases to natural resolution in all. One could also pose the question as to whether months or possibly years of pain, loss of movement, functional restriction and sleep disturbance is likely to be acceptable to the individual affected. Patients with frozen shoulder describe it as an intensely painful and disabling condition that had major negative impact on their quality of life. Jones et al. reported that patients had to leave work, change cars and struggled with the most basic activities of daily living over prolonged time periods. ¹ Pain and disruption were compounded by the lack of visible symptoms and awareness of others resulting in patients reporting feeling their levels of suffering were often unacknowledged. It would therefore not be unreasonable for patients to expect their health care professional to answer what is likely their most pressing question – what is the most effective treatment for frozen shoulder?

The current management of frozen shoulder

The pain, movement restriction, loss of function, disability and associated morbidity experienced with frozen shoulder are such that patients often request intervention. In current practice this tends to be one, or combinations of:

1. Advice, education and pain relief.

2. Physiotherapist-led interventions aiming to increase range of movement and reduce pain – exercises, stretching, joint and/or soft tissue mobilisation.

3. Injection therapy – e.g. intra-articular (IA) corticosteroid injections performed either by anatomical landmarks or under ultrasound guidance, and sub scapular nerve blocks.

4. Manipulation under anaesthetic (MUA) – under general anaesthetic the shoulder is moved through full range of movement in a forceful but controlled manner to tear the joint capsule and restore movement.

5. Arthroscopic capsular release (ACR) – surgical release of the glenohumeral joint capsule to restore movement.

6. Hydrodilatation (HD; also referred to a distension arthrography) – large volume of fluid, typically a combination of saline, anaesthetic and steroid is injected into the glenohumeral joint under radiological guidance. This may or may not be to the point of capsule rupture.

Physiotherapy led interventions and injection therapy

The current research base regarding the optimal treatment for frozen shoulder including exercise, mobilisations, electrotherapy and injection therapy typically relies on small clinical trials of limited methodological quality. A range of systematic reviews over recent years^32–35 consistently conclude that due to the low quality of the evidence, recommendations regarding the clinical superiority of one form of treatment over the other could not reliably be made. When evidence for physiotherapy regimes has been found (albeit weakly) the level of treatment required – an average of three sessions per week over four weeks with relatively small gains in pain and function ³⁶ both the clinical and cost effectiveness of such an approach could be questioned.

Release procedures

Release procedures remain commonplace and for the purposes of this review, release procedures include MUA, ACR and HD. It has been estimated that over 14,000 patients in England undergo release in the form of a capsular release or MUA per year. ³⁷ This figure does not include those who have undergone HD. Despite this prevalence and the plethora of small scale trials, audits or case series that have investigated the outcomes of MUA, ACR or HD^{14,23,28,38,39} good quality comparative studies concerning these procedures are scarce. Maund et al.’s systematic review ³³ including all three forms of release procedures found only one RCT of sufficient quality which compared MUA in combination with home exercises to home exercise alone with a 12-month follow-up. ⁴⁰ Other studies were underpowered, judged to be at high risk of bias, and did not adequately report methods for randomisation or allocation concealment. Studies frequently use different study populations – for example excluding patients with diabetes or not and there is inconsistent use of outcome measures. ³¹ A more recent systematic review and meta-analysis on the effectiveness of HD concluded that the pooled effect size of HD on pain, range of movement and disability was small to insignificant. ⁴¹ As with Maund et al.’s review, the number of studies deemed to be of sufficient methodological quality was small, but Saltychev et al. concluded that the effect sizes following HD generally appear to be very small. At the time of writing, the results of the UK FroST which seek to evaluate the clinical and cost effectiveness of physiotherapy including corticosteroid injections, MUA and ACR were not yet available and will hopefully provide greater clarity regarding the effectiveness of these procedures.

Although of relatively low methodological quality, there has been a considerable volume of prospective studies and case series of patient outcomes following release procedures. A recurrent and interesting theme emerging from these studies is that the early range of movement gained immediately following release procedures for frozen shoulder are not always permanent. Following MUA or ACR around 17% of patients fail to maintain their initial improvement and between 5% and 10% deteriorate to the point where their release would be considered unsuccessful.^14,38 In a single surgical case series of 730 patients a small number of patients failed to improve after repeated MUAs. ³⁸ Three per cent of patients experienced worsening symptoms following their repeated procedure. The regression experienced by patients can be quite profound and is most pronounced between release and six weeks, despite a 10-week exercise programme. Little improvement was seen after 12 weeks. ²⁸

In addition to the group of patients that fail to improve or even regress, a small subgroup of patients undergoing release procedures for frozen shoulder has been identified that had full or significantly improved range of movement when they were anaesthetised, but before the release procedure had been performed. This was found to occur in 13 out of 730 (1.7%) patients in Woods and Loganathan’s series ³⁸ and similar findings were made in a small-scale study of five subjects assessing range of movement in patients with frozen shoulder following anaesthetisation where differences of between 44° and 110° were found in all patients. ⁴² The findings from both studies could be interpreted as an indication that, in this group, true capsular contracture could not have been the driver for symptoms.

In summary, there is limited evidence for the effectiveness of any intervention for frozen shoulder, be it conservative or surgical, and there is no robust evidence regarding the superiority of one treatment over the other. The National Institute of Health and Care Excellence Clinical Knowledge Summary base recommendations on the jointly produced care pathway between the British Elbow and Shoulder Society (BESS) and the British Orthopaedic Association, which suggest that until further evidence becomes available, any treatment should be undertaken as a result of a shared decision-making process based on the level of symptoms and functional limitations of the individual affected. ²⁶

The post-procedural management of frozen shoulder

For patients who do ultimately undergo release procedures, regardless of the intervention chosen, consensus suggests that the post-operative/post-procedure pathway should involve a period of rehabilitation. This is hypothesised to maintain the movement gained during surgery and support patients to return to everyday activities.^6,30,23,43 The post-operative management that results in optimal outcome for patients following release procedures for frozen shoulder has not been investigated and no accepted guidelines or evidence based best practice currently exist. Different hospitals and different surgeons prefer different post-operative regimes. The extent of this variation has never been documented nor have comparisons of outcomes been made. The considerable variation in practice that exists is summarised in Table 1.

Table 1.

Variations in post-operative physiotherapy regimes.

First author (publication year)	Study design	Procedure	Time from procedure to start of rehabilitation	Content of rehabilitation	Duration	Frequency
Kivimäki et al. (2007) 40	RCT	MUA	Not reported	Exercises explained and supervised in two sessions and written instructions for home exercise programme were provided.	Not reported	Not reported
Jenkins et al. (2012) 76	Retrospective case control study	MUA	Within 48 h	Hydrotherapy Progressed to land-based exercises	Until patient and therapist satisfied no further improvement was possible	On average, one hydrotherapy session and three land-based sessions
Sokk et al. (2013) 77	Prospective cohort study	MUA	Immediately on return to ward	Passive ROM Gentle active assisted movement	Not reported	Not reported
Kraal et al. (2017) 78	RCT	MUA	Within 4 h	ROM exercises Joint mobilisations Maitland grade 3, 4 & 5 Home exercises Stretching – end of range for 2 min	Not reported	Week 1 Daily visits Week 2 + If ROM is maintained – general exercise programme to regain function of rotator cuff and scapula muscles
Woods and Loganathan (2017) 38	Prospective case series	MUA	Day after procedure	Pendular exercises Wall climbing movements	Not reported	Not reported
Smith et al. (2014) 14	RCT	ACR + MUA	Not stated	ROM exercises	Monthly	3 months
Bidwai et al. (2016) 30	Prospective case series	ACR + MUA	Day of procedure	Maintenance of ROM with emphasis on external rotation	Not reported	Daily in the first week
Le Lievre and Murrell (2012) 79	Retrospective case series	ACR	Day after procedure	Week 1 1st session – PROM & AROM 2nd session – 10 repetitions of active assisted ER exercises with a broom handle. Week 2 Resisted ER, IR and adduction with theraband Free weight resistance flexion & abduction. 10 s stretches, repeated 5 times of flexion, horizontal adduction and external rotation in 0° and 90° of abduction. Rotator cuff muscle strengthening	10 weeks	Supervised sessions – twice a week in week 1 Home exercise programme – every 2 h 3 sets per day
Ebrahimzadeh et al. (2014) 80	Prospective case series	ACR	Day after procedure	Aggressive physiotherapy	Not reported	Not reported
Barnes et al. (2016) 28	Prospective case series	ACR	Day after procedure	Strict and intense post-operative rehabilitation regime. Week 1 – Patients were instructed to perform the exercises every 2 h. 2nd physiotherapist visit – 10 repetitions of active assisted external rotation movements using a broom handle. 2nd week, patients were required to perform 3 sets of 10 exercises a day using theraband including rotator cuff strengthening	10 weeks	Every 2 h in first week Daily for 10 weeks
Buchbinder et al. (2004) 81	RCT	HD	Not reported	Pendular exercises and scapular setting	Not reported	Not reported
Mun and Baek (2016) 43	RCT	HD	Immediately	Self-exercise programme booklet	Not reported	Twice per week
Sinha et al. (2017) 23	Prospective case series	HD	Same day	5 exercises including flexion, abduction and rotation	3 months	Not reported
Robinson et al. (2017) 74	RCT	HD	Same day	Home exercise programme Wall slides (flexion/abduction) AAROM exercises (flexion/abduction/IR/ER) Passive stretches Scapula setting Isometric strengthening Heat pack Supervised physiotherapy Manual therapy Therapist applied stretches Glenohumeral joint mobilisations Cervical & thoracic spine mobilisations	4 weeks	5–10 repetitions 5 times per day Weekly
Rangan et al. (2015) 26	Care pathway	MUA/ACR/HD	Prompt	Not reported	12 weeks of physiotherapy required to maintain range of motion	Not reported
Brealey et al. (2017) 82	UK FroST Trial protocol	MUA/ACR/HD	Within 24 h	Advice and education Supervised and home exercises Strengthening, sustained stretching Manual shoulder mobilisation Soft-tissue techniques PNF Spinal/scapulothoracic manual therapy Posture correction Discouraged treatments: Bowen therapy, electroacupuncture, graded motor imagery, mirror therapy, SWD, ultrasound. Prohibited treatments: Brace, craniosacral therapy, deep friction, interferential, laser, shockwave therapy Treatment to be given by a qualified physiotherapist (not a student or assistant)	12 weeks	At the discretion of treating clinician

ACR: arthroscopic capsular release; HD: hydrodilatation; MUA: manipulation under anaesthetic; RCT: randomised controlled trial; ROM: range of motion.

The information presented in Table 1 demonstrates a strong preference across studies towards starting post-operative/post-procedural rehabilitation as soon as possible and normally within 48 h. An interesting exception was found in a RCT comparing hydrodistension with joint manipulation under interscalene block with IA corticosteroid injection. ⁴³ The hydrodistension group began rehabilitation immediately following their procedure whereas the IA group waited two weeks – the rationale for which was to allow the pain to settle before the same rehabilitation exercise programme was commenced. As can be seen, there is a physiotherapeutic smorgasbord of treatment modalities employed across variable timescales with inconsistent levels of supervision or contact. Programmes also range from vague to specific regarding frequency and repetition of exercises indicating the lack of clear consensus regarding optimal rehabilitation following release procedures.

Post-release rehabilitation: Are we getting it right?

As briefly indicated, there are potentially different but interconnected mechanisms at play that may determine the course of symptoms in an individual with frozen shoulder. These include (i) inflammation and immune cell response leading to a disruption of connective tissue homeostasis and fibrosis^8,15,44,45 or (ii) potential central pain processes relating to long-standing pain and disability.^46–51 We hypothesise that these causal pathways may be associated with a differing response to treatment (both conservative and surgical options) and may also suggest that post-release rehabilitation may need to take into account the potentially different underlying causes of frozen shoulder.

MUA, ACRs and HD all seek to anatomically release fibrotic tissue, not disrupt the pathological process that leads to that point, i.e. the imbalance in connective tissue homeostasis. If this is a definitive procedure then in theory, maintenance of the range of movement achieved at the point of release should be relatively straightforward to maintain. If, however the disease process remains active and altered cell proliferation and remodelling continues, are some forms of rehabilitation more effective than others in maintaining the aforementioned disruption?

Frozen shoulder is also notoriously painful, and patients have often suffered with intractable pain for many months prior to any release procedure being undertaken. The emerging research presented suggests that potential central pain mechanisms known to be present in long-standing shoulder pain could potentially play a greater role than previously considered.^52,53 The intense and aggressive post-procedural therapy often advocated in the period immediately following release on a background of long-standing inflammation could result in unintended consequences by perpetually driving central pain responses. This may be compounded by the fact that in certain patients, pain perceptions and health behaviours such as fear avoidance, maladaptive coping and having an external locus of control are associated with poor outcome with pain, disability and depression being highly prevalent in frozen shoulder populations.^51,54–56

Treatment selection: Can novel basic science guide us in the future?

Over the past three decades our understanding of frozen shoulder pathophysiology and theories as to what happens have evolved. Despite this there still has not been any clear breakthrough in understanding why it occurs. Most authors now accept that frozen shoulder is a disease characterised by fibroblast proliferation, activation and matrix remodelling (Figure 1). The resulting lesion being the production of a stiff, contracted, diffuse extracellular matrix (ECM) production with the overproduction of densely packed collagen fibrils.^{12,15,57–59} More recently it has been suggested that this fibroblast activity represents a pro-resolving phenotype. ⁶⁰ While previously debated, there is now a relative consensus that this involved immune cell regulation, including mast cells, macrophages, T-cells and B-cells.^12,15

Figure 1.

Schematic representation of shoulder capsule molecular microenvironment in health, following damage and during re-modelling and repair. Following damage, activated fibroblasts, macrophages, aggregated platelets and ECM products direct immune cell activation and tissue repair. These in turn promote T/B-cell chemotaxis which releases a cytokine milieu directing fibroblast proliferation, activation and trans-differentiation, resulting in collagen production and ECM remodelling.

Many studies have identified evidence of elevated cytokine expression, in particular interleukin (IL)-1β, tumour necrosis factor-α (TNF-α) and IL-6 and to a lesser extent IL-8.^61–63 Other authors have identified abnormalities of transforming growth factor β (TGF-β), platelet-derived growth factor-β, cyclooxygenase-2 and vascular endothelial growth factor signalling.^62,64,65 MMP dysregulation has been evidenced by many groups and of note a reduction of MMP-1, increased TIMP-1 and altered MMP1:TIMP-1 ratios.^66,67 Other research has implicated a role for alarmins, neoinnervation and advanced glycation end-product activation.^68,69 There is a clear need for the development of new treatments for frozen shoulder.

There has been a dearth of attempts to develop new therapies in the treatment of frozen shoulder. Some of attempts to treat the disease from an understanding of the basic science include utilisation of anti-TNF-α monoclonal antibodies, collagenase and hyaluronic acid. In a limited trial, one group compared subcutaneous adalimumab injections (anti TNF) to IA glucocorticoid injections and found no difference in outcomes. Given the absence of a true control group they failed to show whether either treatment was effective. ⁷⁰ Badalamente and Wang have conducted a RCT comparing a clostridium histolyticum collagenase injection therapy to exercise-based physiotherapy and demonstrated some evidence of improved pain and function scores compared to exercise alone. ⁷¹ Several groups have trialled hyaluronic acid injections, but the trials were of low quality and results inconclusive. ⁷²

Utilising the knowledge gained from basic science studies on patient tissue may help direct us to treating different molecular subsets of patients with different physical therapy regimes. For example, preliminary evidence suggests that painful stretching may influence molecular tissue homeostasis. Lubis and Lubis ⁶⁷ assessed 50 patients with frozen shoulder who were randomised into two groups – one incorporating exercises which were painless, the other group performing exercises up to what the authors described as “tolerable pain”. The hypothesis being investigated was whether mechanical load could regulate MMP expression and activation. Serum levels of MMPs, TIMPs and TGF-β1 were taken and compared to control subjects. The intensive stretching group demonstrated changes in MMP:TMP ratios that would favour the fibrolytic process and collagen degradation thus restoring normal tissue homeostasis. The authors hypothesised that excessive stress placed on capsular tissue during intensive stretching may influence chondrocyte activity.

Whilst this study was performed in a non-surgical cohort, it does raise the interesting possibility that physical therapy regimes can influence molecular tissue healing towards resolution. Thus, in the future molecular signatures in frozen shoulder patients may be informative to help guide stratified cohorts, possibly based on blood/tissue biomarkers, towards early targeting physiotherapy regimes versus those post-surgical patients who may require adjuncts to physiotherapy treatments early in their rehabilitation.

Is earlier better?

The suggestion from the collective research presented in Table 1 advocate immediate or at least early rehabilitation following release procedures. The rationale suggested is that this is required to maintain the ranges of movement achieved following release procedures to prevent recurrence. Given the nature of the procedures which invariably involve capsular rupture; this is also the time where patients are likely to experience the greatest levels of pain. In a prospective study of 136 patients with frozen shoulder undergoing ACR, Smith et al. ¹⁴ found that between 50% and 80% of patients reported good pain relief between one and six weeks with an average of 16 days to reduce pain on a visual analogue scale from 6.6 down to 1. Other studies have reported a 10-fold reduction in the frequency of extreme pain at one week post-operatively ²⁸ or reported dramatic improvements in pain and function at the two-week point. ⁷³ It cannot be determined whether pain lessens because of the early mobilisation employed in these studies but it could be argued that physiotherapy should start at the point of pain reduction – around one to two weeks following intervention.

If the timing of intervention has received minimal attention in the literature, the nature of post-operative rehabilitation has received even less. The only attempt at comparing post-operative rehabilitation was a study evaluating outcome following HD in 41 patients. ⁷⁴ No difference in Oxford Shoulder Scores was found between those patients undergoing supervised physiotherapy which included advice, exercise therapy, glenohumeral joint, cervical/ thoracic spine mobilisations, glenohumeral passive range of movement, therapist applied stretches and manual therapy to a home exercise programme compared to a self-directed home exercise programme.

We have proposed that the gains in movement achieved are often not maintained despite physiotherapy intervention which may be due to an immunochemical mechanism or potential pain syndromes. Despite common reference to its importance and the general assumption that it is required, there has been limited investigation into the optimum nature and duration of rehabilitation following release procedures. The information presented within this review raises the following questions:

1. Is the rationale for release procedures for frozen shoulder fully understood?

2. Is the variability of outcome due to heterogeneity of patients undergoing release procedures, i.e. they do have differing types or presentations of frozen shoulder, with differing prognosis?

3. Does the variability of post-release rehabilitation affect the outcome of release procedures?

With the level of uncertainty around the pathological drivers of frozen shoulder, it may be that for certain patients, current release procedures and associated rehabilitation are underpinned by fundamentally flawed principles. Patients who are classed as non-responders may respond poorly because the treatments currently on offer do not match their disease. Further research is needed to carefully characterise patients presenting with a recent diagnosis of frozen shoulder and prospectively assess their outcomes in relation to the different potential mechanisms, presenting characteristics, and treatment offered.

Conclusions

Despite the passage of 100 years since frozen shoulder debuted in the literature, we seem to have made frustratingly little progress in understanding the pathology and treatment of this condition. The generally low methodological quality and small size of many clinical trials have hindered the development of a robust evidence base on which clinicians and patients can jointly make sound judgements regarding their care. ⁷⁵ This results in an increased level of disease burden to patients and increased cost to the health economy.

It is increasingly apparent that the spectrum and severity of symptoms experienced by each patient varies greatly. It is unclear whether subgroups of patients with what has been identified as “frozen shoulder” exist and whether disease progression can be reliably altered with intervention. Ideally, patients would be stratified into a clinical pathway which is the most likely to be effective, but there is insufficient evidence regarding frozen shoulder presentations and prognosis to underpin a reliable approach to sub classifying patients based on identified clinical parameters be they structural, immunochemical or patient factors such as lifestyle, beliefs or personality traits. If surgical intervention is deemed appropriate, then the post-release “hit them quick and hit them hard” philosophy has perhaps been given more credence that it deserved given the evidence base and should be challenged with further work. When patients are therapy resistant in the post-release phase, multi-disciplinary pain management strategies may be worthy of consideration.

Ethical Review and Patient Consent

Ethical approval was not sought for this article as a review of literature it was not required.

ORCID iDs

Elaine G Willmore https://orcid.org/0000-0002-7530-0177

Neal L Millar https://orcid.org/0000-0001-9251-9907