Objective: To update an existing systematic review of randomized clinical trials evaluating the clinical effectiveness of non-surgical management interventions for people with primary frozen shoulder in terms of pain, movement, self-reported function and disability, quality of life, recovery time, return to work and recreation, and adverse events.

Data sources: Cochrane CENTRAL, SCI and MEDLINE, CENTRAL between 1 January 2010 and June 2017, plus reference lists of included trials and trial registers. Abstracts were independently screened by 2 reviewers and discussed.

Data extraction: Two reviewers evaluated eligibility. Data were extracted by one reviewer and checked by another. Two reviewers evaluated risk of bias. Meta-analyses were not appropriate. Narrative analyses were performed for trials evaluated as low risk of bias.

Results: Thirty trials were included, with the majority of studies evaluated as being at high risk of potential bias. Only 4 trials were evaluated as being at low risk of bias and this, plus the variety of participants included/excluded in trials and the variety of methods, interventions and outcomes used across the trials provided limited new evidence to inform the non-surgical management and treatment of people with frozen shoulder.

Conclusion: Substantial evidence gaps remain for the non-surgical treatment of people with frozen shoulder.

Accepted Jun 5, 2019; Epub ahead of print Jun 24, 2019

J Rehabil Med 2019; 51: 00–00

Correspondence address: Jeremy Lewis, School of Health and Social Work, College Lane Campus, University of Hertfordshire, Hatfield AL10 9AB, UK. E-mail: jeremy.lewis@londonshoulderclinic.com

Frozen shoulder commonly affects people aged around 50 years and is associated with substantial levels of shoulder pain and stiffness that may last for many years. Many people with frozen shoulder report that simple activities, such as dressing, and washing or drying their hair, become almost impossible. The condition may adversely affect the ability to work and frequently causes severe interruptions to sleep. The reasons why up to 5% of the population develop frozen shoulder are unknown. Many treatments, often lacking sound research evidence, have been recommended. Inappropriate treatment may not resolve the symptoms, may be associated with unnecessary expense, and may even cause harm. We have updated a review of the available literature to synthesize the findings of the available research so that we can make recommendations for the best current treatment alternatives to help people with frozen shoulder and for future research.

Frozen shoulder (FS) is associated with prolonged shoulder disability and is often characterized by severe pain, loss of movement and disrupted sleep (1). The main priority for people experiencing FS is to achieve pain-free freedom of movement as soon as possible, and they are concerned about delays in receiving care and receiving contradictory advice regarding treatment (1). Estimates of incidence of FS range from 0.75% to 5.0% of the population (2, 3) and are higher in people with diabetes (10–46%) (4, 6). Uncertainty remains regarding the distribution of FS between men and women (2), and whether FS is more common in women or is evenly distributed (3). Uncertainty also remains regarding the time course of FS: reports include a mean of 15 months (7), 30 months (8), and 41% of people with symptoms at 52 months (9). One case series (n  =  62) suggested that 50% of people reported mild shoulder pain and stiffness and 60% restricted range of movement at a mean of 84 months (10).

A definitive understanding of the pathogenesis of FS remains elusive (11). Inflammation, fibrosis and contraction of the glenohumeral joint capsule are suggested to explain the symptoms (12) and may be triggered by increased expression of cytokines and neuropeptides (11). However, capsular contraction may not be the only explanation; a small pilot study (n  =  5) by Hollmann et al. (13) reported that, when given a general anaesthetic, people presenting with FS exhibited increased range of movement in shoulder elevation (minimum increase 55°, maximum 110°), suggesting that muscle guarding may partly explain the movement restriction in a percentage of people with FS. FS appears to be most common in people aged in their 50s and 60s (14), and so may relate to genetic (15) or endocrine disorders, such as diabetes and thyroid problems that may become more prevalent at this time (11). FS has been described to occur in 3 transitionary phases: increasing pain and progressive stiffness, ongoing stiffness and decreasing pain, and a resolution phase, in which the remaining pain settles, and movement improves (2, 8, 16). Lewis (17) suggested a simplified method of classification; a pain greater than stiffness phase and a stiffness greater than pain phase. When there is no known reason for onset, such as following surgery, or if the condition is not associated with comorbidities, the term primary (or alternatively idiopathic) FS is recommended (18).

Whether FS is a self-limiting condition that may do well without intervention (7) or whether resolution requires treatment remains unclear (19). A retrospective study suggests that a “no treatment” option may be considered (20): 94% of people (n  =  83, mean time after onset = 9 years) with FS, recovered to normal levels of function and motion without treatment. Management options for FS broadly fall into 3 categories; advice, support and empathy, while allowing natural history to take its course (wait and watch); more formal non-surgical management; and surgical intervention.

A comprehensive, high-quality systematic review of non-surgical management for people with FS concluded that data from studies with a low risk of bias were sparse (21). The following conclusions were drawn from a minority of studies deemed to be of low risk of bias (21): possible short-term benefit from adding a single intra-articular steroid injection to home exercise for patients with primary FS of less than 6 months’ duration, adding physiotherapy (including mobilization in 8–10 sessions over 4 weeks) to a single steroid injection, adding shortwave diathermy (SWD) to passive mobilization and home exercise (for some outcomes only), and high-grade mobilization may be more effective than low-grade mobilization in a population of patients who have already had physiotherapy and/or steroid injection (for some outcomes).

Recommendations to classify FS into diagnostic subcategories have been made, with the aim of developing more homogeneous research investigations and better understanding of whether differences in pathogenesis and management exist within the different subgroups (17). The term primary or idiopathic FS appears to be the most common presentation and is the diagnostic category when there is no identifiable reason for onset. Secondary FS is used when there is an identifiable potential cause preceding the onset of the condition. Secondary systemic FS describes FS in the presence of diabetes. When the condition is preceded by a humeral or clavicular fracture, a chest wall tumour, cervical radiculopathy, ipsilateral breast surgery or cerebrovascular accident, the diagnostic label secondary extrinsic FS has been suggested. FS occurring post-surgery is termed iatrogenic FS.

The current review aims to update the existing review (21) and evaluate the clinical effectiveness of non-surgical management interventions focussing on primary or idiopathic FS, in terms of pain, range of shoulder joint movement, self-reported function and disability, quality of life, recovery time, return to work and recreation and adverse events.

This review was carried out following recommended advice from the Cochrane Handbook, reported according to the PRISMA statement (22) and registered in advance with PROSPERO (reference number: CRD42015013728).

Randomized controlled trials (RCTs) containing participants with primary (idiopathic) FS or its synonyms, such as adhesive capsulitis, were included. FS in people with diabetes has been considered as both primary and secondary. For this review, unlike Maund et al. (21), but in agreement with a consensus on defining the subcategories of FS (18), people with diabetes were considered to have systemic secondary FS and these studies were excluded. Participants with other non-idiopathic or secondary causes of FS (for example trauma or surgery), as well as participants with symptoms indicating an alternative source of shoulder pain (for example, referred pain, paraesthesia, instability, rotator cuff tendinopathy) were excluded. Studies with participants with general shoulder symptoms (shoulder pain or symptoms or non-specific shoulder pain) were not included. Studies needed to state they were RCTs; studies stating participants were “divided”’ or “assigned”’ to groups were not assumed to be RCTs and were not included.

Physiotherapy and rehabilitative interventions, distension, oral medications, and injection procedures were included. Surgical procedures, such as arthroscopic and open capsular releases, were excluded.

Trials comparing an intervention and control group, or, comparing 2 of the included interventions were included. Trials comparing an intervention, a further intervention outside of the review and a control were included only if the data could be extracted for the intervention and control arms.

Outcomes were: adverse events, pain, range of shoulder joint motion, self-reported function and disability, strength, quality of life, recovery time to recreation and return to work.

Electronic searches were initially conducted in December 2014 by 2 independent reviewers (EB, NDB) and subsequently updated by 2 independent reviewers between November 2016 and June 2017 (CML and an experienced healthcare librarian). The original review (21) searched 19 sources, but concluded that 2 databases (Cochrane CENTRAL and either the Science Citation Index or MEDLINE), plus reference checking, proved effective in identifying included papers and this approach was utilized in this review (23). The original MEDLINE search (Table SI) was re-run in OVID (Epub Ahead of Print, In Process & Other Non-Indexed Citations, Ovid MEDLINE (R) Daily and Ovid MEDLINE (R) 1946 to Present) with search date limits 1 January 2010 to 8 December 2016. In addition, Cochrane CENTRAL, SCI MEDLINE, CENTRAL databases were searched, following the search strategies detailed by Maund et al. (21), to retrieve randomized clinical trials published between 1 January 2010 and 6 June 2017. Reference lists of included trials were manually searched for other relevant trials (no new hits). Clinical trials registers were searched to identify any additional publications and to identify on-going trials to further inform the review. ClinicalTrials.gov (searched 8 June 2017, n = 42: no new hits), ISRCTN (searched 7 June 2017, n = 20: no new hits) and the European Union Clinical Trials Register (searched 8 June 2017, n = 3: no new hits). Although resources for translation were not available, searches were not restricted to English language, so that the quantity of non-English-language research could be established.

Two reviewers independently screened the retrieved title and abstracts against the inclusion criteria (EB and NDB in 2014, CML and JL in 2016/17). When potentially relevant studies were identified, or information was insufficient, the full-length article was screened and discussed by 2 reviewers. Two reviewers independently screened the full articles for inclusion (EB and NDB in 2014, CML and JL in 2016/17) and then discussed them together. In the case of disagreement between the 2 reviewers, a third reviewer was consulted (JL in 2014 and NDB in 2016). This occurred in the case of 2 trials in 2014, but was unnecessary in 2016/17.

Risk of bias was assessed by 2 independent reviewers using a domain-based, risk of bias assessment approach (24). Domains included sequence generation, allocation concealment, intervention integrity, effective blinding and whether outcomes were pre-specified, analysed, and reported appropriately. Any additional aspects of study methodology relevant to validity or generalizability were also evaluated (e.g. appropriateness of study measures and sample size). Items relevant to each trial’s internal validity were reviewed and graded as adequate (low risk of bias), partial (moderate risk of bias), inadequate (high risk of bias), or unclear (uncertain risk of bias). Blinding of participants and intervention providers was not thought feasible, and thus lack of blinding here was rated as low risk. Any items judged differently by reviewers were discussed and resolved by 2 reviewers (CML, JL) until consensus was reached and an evaluation of overall risk of bias was achieved.

Study characteristics for all trials included in the review were extracted by one reviewer using a standardized form (CML) and checked against their source article by a second reviewer (JL).

Data syntheses were undertaken for trials evaluated as low risk of bias. There are different methods and approaches towards undertaking systematic reviews, and a lack of consensus regarding the optimal approach researchers should utilize in reviews. Whilst some researchers decide to include all studies, including those at high risk of bias, in syntheses, others consider it to be inappropriate and potentially misleading to include studies at high risk of bias in all parts of data analyses/syntheses within a review (25, 26). It was unclear from the written publication whether participants with diabetes mellitus were included or excluded in trials for 10 trials evaluated at low risk of bias. The corresponding authors for these trials were emailed requesting additional information, with a follow-up email at least 2 weeks later, where necessary. Following this, it was confirmed that 4 trials were eligible for inclusion and 6 were excluded because they included participants with diabetes mellitus or because they did not reply to confirm eligibility. A further author (Ma et al.) (27) was emailed to clarify how participants were randomized; no reply was received, therefore this trial could not be evaluated as low risk of bias.

In the OVID MEDLINE search for trials 2,724 records were identified, 76 records were screened, and 55 full papers were subsequently assessed against the review’s inclusion criteria. Of these, a total of 30 trials were included in the review. No additional studies were identified from searches of the additional databases. A summary is provided in the PRISMA flow diagram in Fig. 1. Table SII¹ contains a list of excluded studies. The characteristics of the trials included in the review and their risk of bias assessments are summarized in Tables I and II. Four studies were included in data syntheses.

Celik & Mutlu (28) compared the effectiveness of joint mobilization and stretching vs stretching alone. Joint mobilization techniques included glenohumeral joint distractions, caudal glides, posterior and anterior glides at a rate of 2–3 oscillations/s for 1–2 min and progressing from Grades I–II to III–IV if pain allowed. The stretching programme involved 20 s of stretching and 10 s of rest and was performed 10 times in the directions of flexion, abduction in the scapular plane, external and internal rotation. Both groups performed a home exercise programme; self-stretching and exercises (scapular retraction, external rotation, extension against resistance, wall and table push-ups, and scapular adduction in prone). All participants received 18 sessions, including the home exercise programme (Table I). This study is a pilot trial, but is reported as a trial, including a sample size calculation and statistical testing, rather than as a pilot trial (29). Both groups improved over time. The sample size (n  =  30) was small and between-group comparisons showed that the effect size for the primary outcome used in the power calculation, the Disabilities of the Arm, Shoulder and Hand (DASH), was 0.03. Although the trend (8 points at the end of treatment and 6 points at 12 months) in the DASH was better in the mobilization and stretching group over the stretching alone group this would not be considered clinically meaningful. The mobilizations and stretching group had improved Constant score and abduction and external rotation range at the end of treatment, and this was maintained at 1-year follow-up. However, it is unclear if the 14° increase in abduction, and 6° increase in external rotation at 1 year is of clinical relevance, although the 17.2 increase in Constant score might be relevant.

Gutierrez Espinosa et al. (30) also evaluated joint mobilization combined with 15 min on an upper extremity cycle ergometer (n  =  29), compared with ultrasound, self-assisted exercises, Codman exercise, Swiss ball exercises and isometric exercises (n  =  28) (Table I). In this study, the participants were positioned in supine in 30–40° of shoulder abduction and external rotation (according to tolerance). A Kaltenborn type III axial distraction was applied, followed by a posterior glide, without oscillations, for 1 min. This was repeated 15 times, with a 1-min rest, and 10 sessions were delivered, 2 or 3 times per week. Pain, range of motion, and function improved with statistical differences in favour of the mobilization and cycle ergometer group (Table I). There was no loss to follow-up.

Only the short-term results were reported (at the end of the 10^th treatment session). The authors reported a mean increase, in favour of the mobilization and ergometer group for the primary outcome measures; passive external rotation of 27° (56.8° vs 30°), passive flexion of 37° (107° vs 69.7°), passive abduction of 22° (70.7° vs 48.8°), and for the secondary outcome measurements; VAS (pain) of 4.4 (0  =  no pain and 10 = worst imaginable pain) in the mobilization group and 5.4 in the ultrasound and exercise group, and, 21 points in the Constant-Murley Score (50.3 vs 29.7).

Although the findings suggest clinical improvement in all the outcomes of interest in favour of the mobilization and ergometer group, the absence of medium- and long-term follow-up is a clear limitation of this study.

Ibrahim et al. (31) compared the effectiveness of a multi-modal treatment programme consisting of heat packs, therapy to facilitate muscle relaxation and glenohumeral mobilizations (inferior glides, longitudinal caudad) for 2 min, using large-amplitude oscillations and repeated 3 times in a 10-min session (n  =  30) and a static progressive stretch device with the multi-modal programme alone (n  =  30). Both groups received treatment 3 times a week for 4 weeks and participants were also provided with a home exercise programme. The static stretch group were asked to apply the static stretch device for a single 30-min session daily for 1 week, twice a day for 2–3 weeks, and 3 30 min sessions per day in week 4. Follow-up assessment occurred at 4, 12, 24 and 52 weeks and at all follow-up time-points the findings favoured the progressive static stretching and multi-modal group for the primary outcome measures of interest. For the primary outcome measures at 1 year there were significant improvements in the progressive static stretching and multi-modal therapy group, for passive external rotation (87° vs 39°), for passive abduction (178° vs 133°), and for active abduction (178° vs 84°). For the secondary outcome measures, the results also favoured the progressive static stretching group for DASH (at 12, 24 and 52 weeks) and for VAS (pain) at 24 and 52 weeks). At 52 weeks the DASH scores were 1.5 vs 55.3 and VAS (pain) were 1.1 vs 3.1 (0  =  no pain and 10  =  worst imaginable pain). The participants were taught how to use the device by the principal investigator, a potential source of bias, and data was collected at a single centre. Compliance data were collected, but not reported.

In a population (n  =  53) of people in the initial pain (freezing) stage of FS Yoon et al. (32) investigated whether a single intra-articular injection of high-dose (4 ml of 10 mg/ml triamcinolone acetonide and 1 ml of 1% lidocaine) (n  =  20) improved pain and function in patients with FS more than a low dose (2 ml of 10 mg/ml triamcinolone acetonide and 3 ml of 1% lidocaine) (n  =  20) or a placebo (5 ml of 1% lidocaine) (n  =  13). Participants were described as having stage 2 of adhesive capsulitis (freezing stage according to Hannafin & Chiaia (16)) with at least one month of pain duration, and mean pain intensity during a day defined as a score of 3 points or more on a 10-cm visual analogue scale rated from 0 (no pain) to 10 (worst imaginable pain). The injections were performed in a hospital setting and under ultrasound guidance. After the procedures the participants were given a home programme that involved stretching, Codman exercises, wall-climbing exercises, external and internal rotations using a bar and posterior shoulder stretching. The exercise programme was to be performed for 10-min, 3-times a day. No other treatments or medications were permitted. Follow-up assessments were performed at 1, 3, 6 and 12 weeks. All groups improved over time. There were significant improvements in favour of the high and low dose corticosteroid groups in comparison to the placebo group but no difference between the high and low dose corticosteroid groups, suggesting that in the short term (12 weeks) corticosteroid has a better effect on pain, range of movement (flexion, abduction, extension, internal and external rotation) and functional outcome (SPADI) than lidocaine alone (Table I). Other than facial flushing (3 in the high-dose group and one in the low-dose group) and dizziness (one in the low-dose and one in the lidocaine group), no serious complications (such as infections) were reported. The authors acknowledged that compliance with the home exercises were not checked and people with higher pain scores than reported in the study may have responded differently to the different doses. The findings suggest that in the short term a single ultrasound guided intra-articular injection of low-dose corticosteroid and a home exercise programme is preferable to high-dose corticosteroid or lidocaine in isolation.

This review has updated the existing 2012 review (21), evaluating the clinical effectiveness of non-surgical management interventions of primary FS in terms of pain, range of shoulder joint movement, self-reported function and disability, quality of life, recovery time, return to work and recreation, and adverse events.

Maund et al. (21) concluded: “Based on a single study (33) (2-arm RCT, quality score 8/13, comparing twice weekly, 30-min sessions of high-grade (Maitland grades III and IV) in the stiff zone, to low-grade (Maitland grades I and II) in the pain-free zone, for a maximum of 12 weeks): “and for some outcomes only, high-grade mobilisation may be more effective than low-grade mobilisation in a population of patients who have already had physiotherapy and/or steroid injection” (21, xv).

The findings of the current review identified 3 new trials, deemed to be of low risk of bias that investigated the use of manual therapy as an intervention. Celik & Kaya Mutlu (28) compared joint mobilization and stretching or stretching alone. The reported improvements in symptoms and range of movement need to be interpreted with caution as they may not have achieved clinically important differences (34, 35).

The uncertainty surrounding clinically meaningful findings also exists for the findings of Gutiérrez Espinoza et al. (30). Whether the improvements reported for the passive ranges of shoulder flexion, abduction and external rotation are clinically important remain unclear. Whilst the improvements in pain and Constant score are encouraging; these findings need to be considered cautiously due to the lack of medium- and long-term follow-up data.

That the addition of a daily static stretching programme plus multi-modal treatment improved range of movement, DASH scores and pain when compared with the multi-modal treatment programme alone, requires further investigation and longer term follow-up. For many healthcare systems, including the National Health Service (NHS) in the UK, the number of treatment sessions and resources included in the intervention may prevent the treatment from becoming widely available.

Maund et al. (21) identified one study that compared high- with low-grade mobilization and reported that, for people who had received a previous corticosteroid injection, the addition of high-grade mobilization may be of benefit. No new study investigating the same parameters was identified in the current review. The 3 new studies deemed to be of low risk of bias in the current review tentatively support the use of manual therapy and stretching in the more stiff than painful stage of the condition (15, 28, 31). However, small sample sizes, uncertainty over clinically important differences, no differences for certain outcome measures, and potential cost vs benefit of the interventions challenges the certainty of any recommendations regarding manual therapy in the management of FS

Two of 6 studies that investigated the use of corticosteroid injections were considered of satisfactory quality in the earlier review (21). In these 2 studies identified concerns were; uncertainty regarding adequate allocation concealment in one study, adequate power in one study, and loss to follow-up, in both studies. In a 4-arm trial, Carette et al. (36) (quality score 9/13) reported best outcomes for a multi-modal treatment that included corticosteroid injections and physiotherapy. In this group, using the SF-36 Physical Component Summary (PCS), the mean score at baseline was 35.2 and the mean change from baseline at 6 weeks was 6.4, at 3 months 8.6, at 6 months 8.8 and at 12 months, 11.5. Also, in this group, the baseline score for the SF-36 Mental Component Score was 43.1, and the mean change at 6 weeks 5.7, at 3 months 6.6, at 6 months 9.2 and at 12 months 9.3. Ryans et al. (31) (quality score 8/13) reported that the mean daytime resting pain score in the group receiving corticosteroid and physiotherapy was 31.2 (out of 100) and at 6 weeks was 14.9. In the injection-only group the score was 28.1; in the placebo injection and physiotherapy group, 28.2; and in the placebo injection-only group, 38.7.

Maund et al. (21) concluded that: “There may be short-term benefit from adding a single intra-articular steroid injection to home exercise for patients with primary FS of  < 6 months’ duration. In the same population there may also be benefit from adding physiotherapy (including mobilization in 8–10 sessions over 4 weeks) to a single steroid injection” (p. xv).”

The current review identified one new trial deemed to be at low risk of bias investigating injection therapy for primary FS (32). Concerns included; short-term follow up, 3 groups, and small sample size. No difference between high- and low-dose corticosteroid groups was identified and both performed better than lidocaine only. Based on their findings, Yoon et al. recommended using low-dose corticosteroid (2 ml of 10 mg/ml triamcinolone acetonide and 3 ml of 1% lidocaine) and the home programme.

As such, the findings of the current review support and extend on those published by Maund et al. (21), and recommend the use of corticosteroid and a home exercise programme for people with a duration of FS symptoms less than 6 months.

Maund et al. (21) concluded: “Based on a single study (3-arm RCT (86) (quality score 7/13) comparing SWD and stretching, heat pack and stretching, and, a home exercise programme), and for some outcomes only, there may be benefit from adding SWD to passive mobilisation and home exercise” (p. xv).

The current review did not identify any new studies that investigated the effectiveness of SWD and, as such, we are not able to add additional information to that provided in the earlier review. In addition, the current review did not find any new research deemed to be at low risk of bias to further inform the use of electrotherapy modalities, acupuncture, taping or dry needling in the management of primary FS.

None of the 4 studies (28, 30–32) deemed to be at low risk of bias reported return to work and/or return to recreation times. Three studies (28, 30, 31) did not report data on the occurrence of any adverse or serious events. Yoon et al. (32) reported no serious complications, such as infections, but they did report 4 cases of facial flushing (3 in the high-dose CS group and 1 in the low-dose group). Two participants experienced dizziness due to vasovagal reactions (one in the low-dose CS group and one in the lidocaine-only group).

Whilst the search strategy was comprehensive, and was developed and performed by the team together with a healthcare librarian, it remains a possibility that other studies exist and have been missed from the review. As the flow diagram shows, the searches returned 7 studies that were not published in English, which could not be fully screened, and so were excluded from the review. Whilst there is conflicting evidence regarding the extent and effects that language bias may have upon review findings (38) it is possible that these studies might have met the review inclusion criteria and provided additional data pertinent to the review, and this is acknowledged.

Discussions between reviewers (CML, JL) following independent risk of bias assessments achieved consensus, and the third reviewer was not required to discuss studies. It was clear that, due to the variety of participants included/excluded in trials, the low number of studies evaluated as being at low risk of bias and the different methods, interventions and outcomes used across the studies meant that it was not appropriate to conduct meta-analyses in this review. Only 4 trials were evaluated as being at low risk of bias and could be included in the data synthesis stage of the review. Limitations were identified in each of these trials, including; small sample size (n = 30) (28), short-term follow-up (30, 32), and potential methodological bias (31).

The aim of this study (21) was to update a previous review and provide guidance to clinicians and patients on any new information deemed to be at a low risk of bias on the non-surgical management of idiopathic FS, which may inform clinical practice and shared decision-making. Unfortunately, the majority of studies were evaluated as being at high risk of potential bias, implying that, despite 30 trials meeting the eligibility criteria and being included in the review, there is limited new evidence to inform the non-surgical management and treatment of people with FS. Given that so many trials were evaluated as being at high risk of bias we believe our decision to exclude these trials from data syntheses to be the correct choice for this review; their overwhelming predominance in syntheses might have led to confusing, inappropriate or misleading findings if they had been included (25, 26). None of the included studies had a no-treatment group, and return to work and recreation data were lacking in all. We recommend that, in addition to reporting adverse events, all future research report these aspects.

In conclusion, there is limited additional guidance available to support clinicians and those seeking care for the non-surgical management of idiopathic FS. An intra-articular corticosteroid injection supported by a home exercise programme may be of benefit for those with symptoms of less than 6 months. In addition, there may be some benefit for including manual therapy and stretching, but due to the high number of treatments required and the uncertainty of achieving clinically meaningful differences, the inclusion of these interventions, must be considered cautiously.

The expertise and assistance of librarian Tatjana Petrinic in developing and undertaking the searches for this review is gratefully acknowledged.

The Health Foundation (http://www.health.org.uk/) is gratefully acknowledged for providing a research grant to support this review.

The authors have no conflicts of interest to declare.