Introduction

Dislocation of the glenohumeral joint refers to the complete loss of contact between the articulating surfaces of the glenoid and humeral head. This diagnosis is usually confirmed by radiography.1 The overall incidence of primary traumatic anterior shoulder dislocation (ASD) in the UK is between 11 and 51 per 100 000 person-years.2–4 The male incidence rate is 2.64 times higher than that of females, at 34.9 per 100 000 person-years.4 Nearly 47.1% of ASD episodes occur between the ages of 20 and 29 years.4 The frequency of instability, therefore, is inversely proportional to the age of an individual with a higher incidence in younger individuals.5 Sixteen per cent of older individuals (ie, over 60 years) sustained an ASD from simple falls6; while ASD in younger individuals (less than 30 years) occurs predominantly in athletic settings.6 ASD can be associated with secondary injuries such as avulsion of glenohumeral ligaments, labral damage, rotator cuff pathologies, axillary nerve injury and bony damage such as the Hill Sachs lesion.7 8 Injury of concomitant structures may predispose an individual to recurrence of dislocation, chronic symptoms, reduced activity participation and decreased quality of life (QOL).9 ASD and recurrence can contribute to shoulder muscle dysfunction and reduced proprioception.10 Indeed, recurrence of ASD is a common complication, especially in young active males with rates as high as 64%.11

Typical principles of ASD rehabilitation include a focus on dynamic strength and control of the glenohumeral and scapulothoracic musculature, proprioceptive retraining and functional progression.12 The deltoid and rotator cuff muscles form a force couple that keep the humeral head centred in the glenoid cavity, while an anterior–posterior force couple is formed by subscapularis anteriorly and the infraspinatus posteriorly.13 Disruption of the deltoid-rotator cuff force couple can give rise to deltoid overactivity, resulting in increased superior translation of the humeral head.14

Exercise-based interventions (EBIs) are an integral component of post-ASD recovery either alone, or in combination with surgical interventions. ASD management requires a phasic, criteria-driven and graded exercise programme that restores strength, range of motion (ROM) and function of the glenohumeral joint.15 The current literature reflects a range of different EBI, as summarised in table 1. There is a need to better understand the effects of EBI (ie, EBI in conjunction with surgery and EBI in the absence of surgery, hereafter: ‘EBI alone’ and ‘multimodal EBI’16 that uses additional strategies other than a home-exercise programme such as neuromuscular exercise and ultrasound-guided elastic resistance training) on recurrence and functional outcomes.17 Neuromuscular exercise in this context includes strength, coordination, balance and proprioception under the guidance of an exercise-based professional; whereas the elastic resistance training uses a range of movements, while using ultrasound to optimise recruitment of appropriate musculature.9 While several studies have investigated EBI as a component of ASD management,9 18 19 to date no systematic review and meta-analysis has evaluated the effectiveness of EBI for ASD by comparing different types of EBI. This study aimed to review and synthesise the literature to compare the effectiveness of surgery in conjunction with EBI to EBI alone on recurrence, return to activity (RTA) and functional outcomes in adults who sustain an ASD in athletic and non-athletic settings.

Table 1

A summary of the multicomponent rehabilitation protocols described in the included studies9 33 34 55 97

Methods

We carried out a systematic review and meta-analysis investigating EBI following ASD. The Preferred Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were followed for the conduct and reporting of this study.20 Further, the PRISMA-S extension21 was used to guide our search methodology. This review was prospectively registered with the International Prospective Register of Systematic Reviews (PROSPERO) (registration number CRD42021262494).

Literature search

We searched the following electronic databases: MEDLINE (PubMed), Web of Science (EBSCO), Scopus (EBSCO) and Google Scholar (Google) for suitable studies conducted from 1990 to May 2022, in English language only. This search was updated in March 2023. Controlled vocabulary (Medical Subject Headings (MeSH) in MEDLINE), for example, “anterior shoulder dislocat*” for “anterior shoulder dislocation”, and “exercise therap*” for “exercise therapy”, were used. The full search strategy can be accessed in online supplemental table S1. From the search, all identified citations were collated and uploaded into Mendeley Reference Manager V.1.19.4 (Elsevier Mendeleyd, London, UK).

Inclusion and exclusion criteria

The ‘PICOS framework’ (ie, population, intervention, comparison, outcomes, study type) was used for the inclusion and exclusion of studies.

Population

Adults who sustained an ASD in an athletic or occupational setting.

Intervention

Any EBI (ie, EBI alone or EBI in conjunction with surgery) for treating ASD, including strength, neuromuscular control/proprioception, plyometrics and mobility training.

Comparison

Standard treatment defined as usual practice either after surgery or as a stand-alone.

Outcomes

Primary outcomes: recurrence, RTA (ie, sport, work or regular activities of daily living).

Secondary outcomes:

Self-reported measures:

American Shoulder and Elbow Surgeons Scale (ASES).

Constant-Murley Score (CMS).

The Disabilities of the Arm, Shoulder and Hand questionnaire (DASH).

Rowe score.

Visual Analogue Scale (VAS).

Western Ontario Shoulder Index (WOSI).

Shoulder muscle strength.

Shoulder ROM.

Study type

We included randomised controlled trials (RCTs), quasi-RCTs or observational studies that only evaluated the efficacy of an EBI either postoperatively, or as stand-alone non-surgical management for ASD. We excluded case reports, secondary research and conference papers. For our meta-analyses, only studies that included both a control and an experimental (ie, surgery+EBI or multimodal EBI) group were synthesised.

Exclusion criteria

Non-human or cadaveric studies.

Studies involving non-ASD shoulder injuries such as recurrent shoulder dislocation, multidirectional instability, shoulder impingement syndromes, acromioclavicular or sternoclavicular injuries, and posterior shoulder dislocation;

Passive interventions or non-EBI such as immobilisation, closed reduction and passive pain modulating physiotherapeutic modalities such as electrotherapy.

Studies that did not describe or report any rehabilitation following an ASD.

Studies that focused on postoperative complications.

Studies that focused only on assessment of traumatic ASD only.

Study selection

Two authors (VC and TL) independently screened the title and abstracts of the studies identified from the search, and then independently screened full-texts of relevant studies. Following this, a reference list search was performed by CD to identify any additional relevant studies. Initially, any disagreements were discussed with a fourth author (RMJdZ) at both abstract and full-text stages, until consensus was reached.

Risk of bias assessment

Two independent reviewers (RMJdZ and VC) assessed the risk of bias and methodological quality of eligible articles using the previously validated Downs and Black checklist.22 Twenty-seven items were rated as yes (=1) or no/unable to determine (=0), and one item (number 27 that assessed power calculation) was rated on a 3-point scale (yes=2, partial=1 and no=0).23 Scores range from 0 to 28 including the adjustment question 27 to a binary (yes/no) response (ie, sufficient power with sample size or not). The higher the score, the better the methodological quality and hence lower risk of bias. The quality of studies was categorised as follows: excellent (26–28), good (20–25), fair (15–19) and poor (14).24 Points were only awarded if a study clearly met the criteria. If there was disagreement between reviewers (RMJdZ and VC), a third assessor (TL) provided consensus.

Data extraction

Data were extracted by TL and CD using a standardised data extraction tool (JBI SUMARI, JBI Adelaide, SA, Australia) and Excel spreadsheets (Microsoft Excel V.2016, Microsoft, Redmond, Washington, USA). The extracted data included details of the population, study methods, interventions and outcomes relevant to the review objective. The authors of papers were contacted to request any missing or additional data, where required. If the authors did not respond within 2 weeks, they were contacted again to follow-up. If data were not obtained within 4 weeks the study, or the relevant section thereof, was not included in the review.

Statistical analyses

Means, SD and sample sizes were extracted for all continuous outcome measures; hedges’ g effect sizes and the respective 95% CIs were calculated, with the magnitude of effect defined using standardised conventions, with small, moderate and large effect sizes aligning with 0.20, 0.50 and 0.80.25 Data were analysed via a change score from premeasurement to postmeasurement using weighted mean differences and Hedges’ g effect sizes in the random effects model (Knapp-Hartung SEs using the Sidik-Jonkman model, as the best model accounting for normality and sparse data bias.26–28 For categorical data (recurrence and RTA), data were analysed using both standardised (risk ratio) and unstandardised models (log risk ratio). For alternative methods of data reporting, they were converted into a corresponding effect size (eg, SE of the mean was converted to SD using the following formula SE x the square root of the number of participants=SD). If data extraction of an included study was not possible, the study was excluded from quantitative analysis. If requested data were not provided, the outcome was excluded from quantitative analysis, but used to inform qualitative synthesis. If data could be obtained from figures or graphs, extrapolation of the mean and respective measure of variance was conducted using digitisation software (Get Data Graph Digitizer), and conversions applied to estimate the respective effect size and 95% CIs. Statistical heterogeneity was investigated for studies by calculating Cochrane’s Q, where significant heterogeneity was indicated by p≤0.10. The magnitude of statistically significant heterogeneity was determined using the I2 statistic, where values of 25%, 25%–75% and 75% represent low, moderate and high levels of heterogeneity, respectively.29 Where heterogeneity exceeded moderate (>50%), follow-up analyses were conducted to investigate the source of this heterogeneity, such as time since surgery, point of measurement (follow-up time). Specifically, a leave-one-out sensitivity analysis was conducted, where the overall effect from removing a single study was examined. All analyses were carried out in Stata V.17.0 MP (StataCorp).

For the primary outcomes (recurrence and RTA), the potential of non-reporting bias was evaluated by using the Outcome Reporting Bias in Trials (ORBIT) framework30 31 to investigate potential missing results. The Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach was used to determine the certainty and strength of evidence carried out in accordance with set recommendations.25 For example, observational studies were assigned a ‘low’ certainty of recommendation prior to then either being upgraded or downgraded from this point, based on the quality of the evidence.32 Studies were upgraded for factors such as large effect sizes or dose–response relationships between the intervention (eg, surgical+EBI) and outcomes (ie, recurrence, RTA, self-report measures and functional outcomes). Studies were downgraded according to GRADE for non-reporting bias, indirect relationships with results (unexplained confounding) or inconsistencies between studies. From this process, qualitative ratings for the certainty of evidence and recommendations were listed as ‘high’, ‘moderate’, ‘low’ or ‘very low’, and were able to be interpreted according to the GRADE approach.32

Equity, diversity and inclusion statement

Our research team comprises female and male members, early-career and mid-career researchers, representation from diverse disciplines, and hailing from three countries. In our study, we specifically focused on individuals with traumatic ASD in both athletic and non-athletic contexts, ensuring representation of both males and females. It is important to note that the majority of studies conducted in this field are carried out in higher resource countries, often published by more developed nations. Unfortunately, there is a notable absence of publications from lower resourced countries, which highlights a disparity in research findings between settings with varying resource levels. We acknowledge that the findings derived from our study may have limited generalisability due to the specific to settings with fewer resources, presenting distinct challenges and requiring tailored approaches. There is a need for further investigations conducted in diverse settings to help bridge the existing knowledge gaps and promote equitable healthcare practices globally. Despite these constraints, our study provides valuable insights that can significantly contribute to future research, interventions and policy-making within the identified context.

Results

Across four databases, 3616 studies were identified, and 842 duplicates were removed. Figure 1 illustrates the identification of studies and each of the stages for the review process according to the PRISMA flow diagram.20 From a total of 2285 titles and abstracts screened, and 576 full-text articles assessed, 60 studies (n=3598) were included for qualitative synthesis and 7 (n=345) studies were included in the meta-analysis. The mean age of participants in the included studies was 26.71±9.19 and 56% of those included were male.

Figure 1

PRISMA flow diagram20 delineating the number of studies included and excluded through each stage of screening. *Indicates the four outlined databases at the top of this figure. **Indicates total number of records excluded at title/abstract stage. ASD, anterior shoulder dislocation; PRISMA, Preferred Items for Systematic Reviews and Meta-Analyses.

Study characteristics

The characteristics of the studies, participants and interventions are summarised in online supplemental table S2. Of the included studies, 46 focused on EBI combined with surgical intervention, and 14 focused on EBI alone. Data from nine studies were pooled for meta-analysis, with a total of 411 participants (n=228 EBI in conjunction with surgery; n=168 EBI alone). Some of the common components of EBI were: mobility exercises such as active ROM, active assisted ROM, capsular stretching exercises and strength training, including: resisted rotator cuff and scapular stabilisation exercises (60 studies). Other interventions were sports-specific training (seven studies), overhead training (three studies), aerobic conditioning (four studies), isokinetic exercises (one study), Bodyblade resisted vibration training (one study), hydrotherapy (one study), plyometric training (three studies) and neuromuscular training (three studies). The median length of follow-up across the studies was 0.9 years and ranged from 3 weeks to 7 years. The included studies used a range of both self-reported and objective measures to assess physical function. In 36 out of 60 studies, the number of male participants was higher than female participants. Overall, there were 2150 male participants and 463 female participants, across the 60 studies.

Quality of studies

Of the 60 studies included in the systematic review, 29 were fair quality (48.3%), 15 studies were good quality (25%) and 16 studies were poor quality (26.7%). The risk of bias (internal validity—confounding bias) within the included studies was low or unclear as a majority of the studies were observational in design (34/60). The individual scoring of each included study can be found in online supplemental table S2. The ‘reporting’ of the included papers scored high, with the commonly unreported aspect being failure to report adverse events following the intervention, and the characteristics of the participants lost due to adverse events. While these studies may not have had participants who experienced adverse events, reporting this in the paper would still have been beneficial. External validity was generally high due to the inpatient, outpatient or home settings incorporated in the included studies. However, all studies restricted normal daily free-living activities until 4–6 months following ASD to enhance internal validity. The included studies took a varied approach to the data collection and analysis of their outcome measures, influencing internal validity. One study used a blinded interpretation framework to reduce the bias of interpretation,9 thereby increasing its internal validity. Three studies blinded the investigator measuring the outcomes,33–35 while many studies (33/60) did not. The level of agreement (kappa) for the methodological quality assessment was 0.90 (weighted kappa: 0.66).

The GRADE certainty of evidence was low for recurrence and RTA (see online supplemental table S3).

Meta-analysis

For each of the primary and secondary outcome measures, the results of the meta-analysis are presented below.

Non-recurrence

The meta-analysis for non-recurrence was based on pooled data from four studies with a total of 216 participants. Of the total, 94 underwent EBI alone while the remaining 122 underwent EBI in conjunction with arthroscopic surgery. There was substantial heterogeneity in the true outcomes of recurrence (I2=51.17%). Non-recurrence was significantly better following EBI in conjunction with surgery. Individuals who underwent EBI in conjunction with surgery were 2.03 times more likely to have treatment success (ie, not sustain a recurrent ASD) than individuals who underwent EBI alone (RR 2.03, 95% CI 1.03 to 3.97) (figure 2) (online supplemental figure 1).

Figure 2

Forest plot from the meta-analysis showing head-to-head comparison between non-recurrence outcomes following EBI in conjunction with surgery and EBI alone. Event, successful treatment; yes, successful treatment with no episode of ASD recurrence; no, unsuccessful treatment marked by episode of ASD recurrence. ASD, anterior shoulder dislocation; EBI, exercise-based intervention.

Return to activity

The meta-analysis for RTA included three studies, with a total of 143 participants. Of the total, 39 participants underwent EBI alone and the remaining 104 underwent EBI in conjunction with surgery. There was low heterogeneity in the true outcomes of RTA (I2=12.68%). RTA was significantly better following EBI in conjunction with surgery. Individuals who underwent EBI in conjunction with surgery appeared 1.81 times more likely to RTA following ASD than individuals who underwent EBI alone, although results ranged from no improvement in RTA to over three times more likely with surgery (RR 1.81, 95% CI 0.96 to 3.43) (figure 3) (online supplemental figure 2).

Figure 3

Forest plot from the meta-analysis showing head-to-head comparison between RTA outcomes following EBI in conjunction with surgery and EBI alone. EBI, exercise-based intervention; RTA, return to activity.

Self report measures

The meta-analysis included the following comparisons:

Rowe score

One study with a total of 65 participants. Of the total, 38 participants underwent EBI in conjunction with surgery and 27 underwent EBI alone (figure 4). The outcomes were in favour of EBI in conjunction with surgery when compared with EBI alone but were not statistically significant (Hedges’ g 0.33, 95% CI −0.16 to 0.82, p=0.19).

Figure 4

Forest plot from the meta-analysis showing comparison between self-report outcomes (ASES, CMS and Rowe Scores) following EBI in conjunction with surgery and EBI alone. ASES, American Shoulder and Elbow Surgeons Scale; CMS, Constant-Murley Score; EBI, exercise-based intervention.

Constant-Murley Score

One study with a total of 30 participants. Of the total, 15 participants underwent EBI in conjunction with surgery and 15 underwent EBI alone (figure 4) (online supplemental figure 3). The outcomes were in favour of EBI in conjunction with surgery when compared with EBI alone and were statistically significant (Hedges’ g 1.60, 95% CI 0.79 to 2.40, p<0.001).

American Shoulder and Elbow Surgeons Scale

One study with a total of 45 participants. Of the total, 23 participants underwent EBI in conjunction with surgery while 22 underwent EBI alone (figure 4). The outcomes were in favour of EBI in conjunction with surgery when compared with EBI alone but were not statistically significant (standard mean difference (SMD) 0.21, 95% CI −0.37 to 0.78, p=0.48).

Overall, for these self-report measures the outcomes were in favour of EBI in conjunction with surgery when compared with EBI alone, but was not statistically significant (Hedge’s g 0.66, 95% CI −1.19 to 2.52, p=0.26). Heterogeneity was I2=82.07% (considerable heterogeneity) for this model (Sidik-Jonkman). The estimates and prediction intervals are shown in figure 4.

We conducted a subgroup analysis for the CMS per the function, pain, ROM, strength and overall composite score components. The scores for each component were compared between the EBI in conjunction with surgery group, and EBI alone group. Heterogeneity was I2=48.35% (considerable heterogeneity) for this model (Sidik-Jonkman). The CMS scores were not normally distributed as indicated by the skewness 1.81 for the mean change score for surgical and EBI and 0.97 for the mean change-score for EBI alone. The subgroup analysis demonstrated the following CMS component specific results:

Pain—outcomes in favour of EBI in conjunction with surgery when compared with EBI alone (Hedge’s g 0.95, 95% CI 0.21 to 1.69).

Function—outcomes in favour of EBI in conjunction with surgery when compared with EBI alone (Hedge’s g 1.43, 95% CI 0.65 to 2.22).

ROM—outcomes in favour of EBI in conjunction with surgery when compared with EBI alone (Hedge’s g 1.14, 95% CI 0.39 to 1.89).

Strength—outcomes in favour of EBI alone when compared with EBI in conjunction with surgery (Hedge’s g 2.32, 95% CI 1.41 to 3.24).

Total score—outcomes in favour of EBI in conjunction with surgery when compared with EBI alone (Hedge’s g 1.60, 95% CI 0.79 to 2.40).

Overall, for this one study, there were improved CMS scores for each subscale and total score for surgical and EBI, when compared with EBI alone (Hedge’s g 1.45, 95% CI 0.82 to 2.09, p<0.001). The estimates and prediction intervals are shown in figure 5 (online supplemental figure 4).

Figure 5

Forest plot from the meta-analysis showing head-to-head comparison for CMS-specific outcomes following EBI in conjunction with surgery and EBI alone. CMS,Constant-Murley Score; EBI, exercise-based intervention.

Physical function measures: strength

The meta-analysis for strength outcomes was based on one study with a total of 45 participants.36 Of the 55 participants, 23 underwent EBI in conjunction with surgery and 22 underwent EBI alone. All participants followed the same rehabilitation protocol for the first 2 months, which included immobilisation in a suspension sling for 3 weeks, and light household activities. The experimental intervention was started 2 months after the surgery. All follow-up assessments (including control group and exercise group participants) were performed individually. The first follow-up was completed 2 weeks after starting the experimental intervention (ie, 10 weeks postsurgery). The second follow-up was 6 weeks after starting the experimental intervention, with subsequent follow-ups at 4 and 6 months after starting the experimental intervention.

Outcomes for each group of muscles tested were compared, as follows:

Grip strength—outcomes in favour of EBI alone, but not statistically significant (Hedge’s g −0.26, 95% CI −0.83 to 0.32, p=0.38);

External rotation strength—outcomes in favour of EBI alone, however, but not statistically significant (Hedge’s g 0, 95% CI −0.57 to 0.57, p=1.00).

Internal rotation strength—outcomes in favour of EBI alone (Hedge’s g −1.05, 95% CI −1.67 to −0.44).

Overall, strength outcomes were not statistically significant (Hedge’s g −0.43, 95% CI −1.78 to 0.93, p=0.31). The heterogeneity was moderate with I2=69.48% (figure 6) (online supplemental figure 5).

Figure 6

Forest plot from the meta-analysis showing head-to-head comparisons between strength outcomes following EBI in conjunction with surgery and EBI alone. EBI, exercise-based intervention.

Physical function: ROMActive AROM

Of the total 75 participants, 38 underwent EBI in conjunction with surgery and 37 underwent EBI alone (figure 7) (online supplemental figure 6).

Figure 7

Forest plot from the meta-analysis showing head-to-head comparison between AROM outcomes following EBI in conjunction with surgery and EBI alone. AROM, active range of motion; EBI, exercise-based intervention.

The meta-analysis for Forward flexion AROM was based on two studies with a total of 75 participants. Of the total, 38 participants underwent EBI in conjunction with surgery and 37 underwent EBI alone. The outcomes were in favour of EBI in conjunction with surgery, but were not statistically significant (Hedge’s g 0.32, 95% CI −0.12 to 0.77, p=0.16). Outcomes for the movements tested for AROM were compared, as follows:

Abduction AROM—The meta-analysis was based on one study with a total of 30 participants. Of the total, 15 participants underwent EBI in conjunction with surgery and 15 underwent EBI alone. The outcomes were in favour of EBI alone, but not significant (Hedge’s g −2.27, 95% CI −3.18 to −1.37).

External rotation AROM: The meta-analysis was based on one study with a total of 45 participants. Of the total, 23 participants underwent EBI in conjunction with surgery and 22 underwent EBI alone. The outcomes were in favour of EBI alone but not statistically significant (SMD −0.40, 95% CI −0.99 to 0.18, p=0.17).

Internal rotation AROM: The meta-analysis was based on one study with a total of 45 participants. The outcomes were in favour of EBI alone but not statistically significant (Hedge’s g −0.14, 95% CI −0.71 to 0.44, p=0.64).

Overall, AROM outcomes were in favour of EBI alone, but were not statistically significant (Hedge’s g −0.40, 95% CI −1.70 to 0.89, p=0.44). The heterogeneity was high with I2=89.68% (figure 7) (online supplemental figure 6).

Passive range of motion

Overall, the passive range of motion (PROM) outcomes were in favour of EBI in conjunction with surgery when compared with EBI alone. However, the outcomes were not statistically significant (Hedge’s g 0.07, 95% CI −0.74 to 0.89, p=0.86). Outcomes for the movements tested for PROM were compared, as follows:

Abduction PROM—The meta-analysis was based on one study with a total of 30 participants. Of the total, 15 participants underwent EBI in conjunction with surgery and 15 underwent EBI alone. The outcomes were in favour of EBI alone, and were statistically significant (Hedge’s g −1.13, 95% CI −1.89 to −0.38, p<0.001).

External rotation PROM—The meta-analysis was based on two studies with a total of 75 participants. Of the total, 38 participants underwent EBI in conjunction with surgery and 37 underwent EBI alone. The outcomes were in favour of EBI in conjunction with surgery, and were statistically significant (Hedge’s g 0.89, 95% CI 0.29 to 1.49, p<0.001).

Forward flexion PROM—The meta-analysis was based on one study with a total of 30 participants. Of the total, 15 participants underwent EBI in conjunction with surgery and 15 underwent EBI alone. The outcomes were in favour of EBI in conjunction with surgery, but were not statistically significant (Hedge’s g 0.19, 95% CI −0.51 to 0.89, p=0.59).

Internal rotation PROM: The meta-analysis was based on two studies with a total of 75 participants. Of the total, 38 participants underwent EBI in conjunction with surgery and 37 underwent EBI alone. The outcomes were in favour of EBI in conjunction with surgery, but were not statistically significant (Hedge’s g 0.27, 95% CI −0.31 to 0.84, p=0.37).

The heterogeneity was high with I2=84% (figure 8) (online supplemental figure 7).

Figure 8

Forest plot from the meta-analysis showing head-to-head comparison between PROM outcomes following EBI in conjunction with surgery and EBI alone. EBI, exercise-based intervention; PROM, passive range of motion.

Qualitative synthesis

Primary outcomes:

Recurrence

Eleven included studies evaluated recurrence following ASD.37–47 In one study,37 the mean (±SD) age of participants was 23.49±7.3 years, with a recurrence rate of 18.2% following an arthroscopic repair and EBI. Another study43 included 20 naval officers who underwent a 4-month EBI, of whom five sustained a recurrence (three of these occurring within 6 months of the ASD). This suggests that clearance of participants to unrestricted sporting or occupational activities must be cautiously recommended during the first 6 months following an initial episode of ASD. Another study44 followed 30 athletes who sustained an ASD. All participants underwent the same course of physical therapy that included strength and mobility training.

On average, there were 1.4 in-season recurrences per season, per athlete. The chances of recurrence were higher in athletes who returned to sport within the same season, an important consideration for return to play decisions. A case series of 42 consecutive patients38—who, on average, participated in sports for 2 hours a day, 3 days a week before they sustained an anterior-inferior shoulder dislocation—reported an overall recurrence rate of 22.5%, mostly occurring within the first year following arthroscopic capsulo-labral reconstruction. All of these except one were contact or overheard athletes. Overhead athletes were more at risk of recurrence compared with other included participants. A negative relationship between the University of California at Los Angeles Shoulder Score (UCLA) and rate of recurrence was observed: that is, lower functional rating on UCLA was accompanied by higher rates of recurrence (33.1% in the low UCLA group vs 29.1% in the high UCLA group). Overall, recurrence was higher within a year of the initial episode in young active males, and in individuals who returned to their preinjury level of activity within 6 months.

Return to activity

Of the included studies, 12 looked at RTA following ASD.9 39 41–46 48–51 The studies used a range of EBIs including postsurgical EBI,39 41 42 45 46 48–51 multimodal EBI9 and EBI alone.43 44 48 50 The majority of studies among non-athletes reported RTA at 3–4 months following ASD, whereas one study of athletes51 reported an average RTA of 8.4 months following arthroscopic stabilisation and postoperative EBI. Most of the included studies report successful RTA (as high as 80%–90%) indicating that EBI can help to facilitate successful RTA. As outlined above, this finding is supported mostly by studies (n=9) that included surgery and postoperative EBI.

Secondary outcomes: self-report measuresRowe score

The Rowe score (0–100; where higher scores reflect greater function) was reported in seven studies,37–40 52–54 of which four37 39 52 54 reported follow-up scores. Studies reported Rowe scores after 13 years37 (median±SD= 90.0±20.5), 6 weeks54 (mean±SD=81.8±24.9 and 84.8±23.3 for dominant and non-dominant), and 2 years (mean=96.5). Archetti Netto et al 52 compared outcomes of open and arthroscopic Bankart repairs (each with postsurgical EBI), and reported the following outcomes: 79% excellent, 14% good and 7% fair Rowe scores (descriptive data not provided). Three studies reported changes from surgery plus EBI from baseline to follow-up including statistically significant improvements (all p≤0.001) in mean scores from preoperative (range: 24–64) to postoperative (range: 80–90).38 40 53 Interventions for each included: arthroscopic Bankart repair plus EBI,38 open reconstruction plus EBI40 and open Bankart repair plus multimodal EBI.53

Western Ontario, Shoulder Instability Index

The WOSI was used as a QOL outcome measure in six studies with 446 participants.5 9 31 42 55 The WOSI is a 21-item scale assessing physical symptoms, sport/recreation/work function, lifestyle function and emotional function (higher score indicates worse QOL). Participants in one study55 demonstrated an improvement in their WOSI scores (mean±SD), 4.5±2.5 years following shoulder reconstruction. They underwent a 4-week postoperative rehabilitation programme (ROM exercises, dynamic strengthening) and returned to activity at 6 months, with a 15%±15% improvement in their WOSI scores at follow-up. Participants in another study9 underwent two different forms of EBI (neuromuscular rehabilitation, or home-based exercises). This study found no statistically significant difference in total or subdomain WOSI scores. Another study among athletes (n=62)56 investigated enhanced-EBI (ie, functional rehabilitation programme comprising supervised ROM, strengthening and plyometric exercises) following arthroscopic Bankart repair. Mean WOSI scores preoperatively were 1578.0±60.9 and 178.9±32.3 postoperatively at 2 years (Δ 1399.1±63.2, p<0.001).

One study57 investigated outcomes cross-sectionally following three to 6 weeks of enhanced-EBI only for both ASD and recurrent-ASD (most-recent occurrence). Those who had experienced an ASD (n=34) had a mean±SD WOSI score of 1064±373.2, compared with 1048.3±371.5 among individuals with recurrent-ASD (n=22). Though no comparison was drawn between a control, or EBI in conjunction with surgery, this study demonstrated that there was no statistically significant between group difference. A second study58 investigated the effectiveness of three EBI only protocols: traditional (ie, resistance band-based exercise), Bodyblade (ie, resisted vibration exercise), and a mixed programme of both—over an 8-week period. All three groups improved significantly at 8 weeks follow-up, with no significant difference between them. At the 8-week follow-up a 59.4% improvement in WOSI score was observed in the traditional EBI group, a 56.5% improvement in the Bodyblade group, and 43.3% improvement in the mixed group. A third study,42 compared two types of surgical management for ASD (traditional vs immediate arthroscopic stabilisation). One group underwent a 3-week immobilisation period followed by physiotherapy, while another group underwent a 4-month rehabilitation programme. All participants demonstrated a significant improvement in WOSI scores at the 24-month follow-up with the immediate surgical group showing significantly better results (287.01±290.19) than the traditional group (633.93±547.25) (p=0.03). In another study,5 252 participants underwent sling immobilisation and a 12-week rehabilitation programme. There was no significant improvement in WOSI score at the 1-year and 2-year follow-up assessments. Overall, the included studies reported on a range of ASD management approaches, and had varying lengths of follow-up and rehabilitation programmes. WOSI scores improved in participants who underwent a well-structured EBI, in conjunction with surgery.

Constant Murley Score

The Constant Murley Score (0–100; where higher scores reflect greater function) was reported in two studies.37 57 This study reports on a 13-year follow-up of patients who underwent arthroscopic Bankart repair with postoperative EBI. Except for a standardised postoperative ROM protocol, a physiotherapist determined an appropriate rehabilitation programme. While no preintervention data are reported, the Constant Murley Score (mean±SD) of 104 participants was 94.0±9.1 at 13-year follow-up, similar to those not reporting recurrence. The Constant Murley Score was also used in the aforementioned study57 cross-sectionally comparing 3–6 weeks of enhanced-EBI for ASD (initial occurrence) and recurrent-ASD (most recent occurrence), respectively. Scores were 70.4±19.4 for the recurrent-ASD group, and 64.4±19.1 for the ASD group. Again, no significant between group difference was observed.

American Shoulder and Elbow Surgeon’s shoulder score

The ASES was used to measure postoperative QOL in six studies with a total of 336 participants.36 37 53 55 59 60 With 100 maximum points, this scale weighs 50% of its questions to assess pain and 50% to assess function. Participants from two studies with a 1-year follow-up reported an improvement of 16 points (95% CI 10 to 23)36; and (64±19.7 to 92.1±3.5 at follow-up p<0.001), following arthroscopic Bankart repair with multimodal EBI.53 Similarly, 83 participants from a study,60 demonstrated an improvement in ASES scores at 33 months following arthroscopic Bankart repair plus multimodal EBI 75.4±17.6 to 94.9±9.6). Another study investigating multimodal-EBI following arthroscopic Bankart repair56 reported significant ASES score improvements. Preoperatively ASES scores were 45.5±3.4, and 2 years postoperatively were 89.3±3.2 (Δ 43.8±4.0, p<0.001). Participants in one study55 reported their satisfaction level (with respect to pain and function) as extremely satisfied following capsule repair (92±12, range: 60–100), with a mean follow-up period of 4.5 years. Another study37 reported ASES scores 93±17.6, but with a longer follow-up period of 13 years following arthroscopic Bankart repair plus EBI. Further, participants who experienced a recurrent episode of shoulder instability during the follow-up period reported lower ASES scores (87.9±15.9) when compared with participants who did not have a recurrent episode of shoulder instability (93.2±9.1). Overall, ASES outcomes were observed to improve through surgical intervention in combination with EBI. All the studies reported an improvement that exceeded the minimal clinically important difference range of 6.4–17 points.61

Disabilities of the Arm, Shoulder and Hand questionnaire

The DASH outcome measure (0–100; 0=no disability; 100=severe disability) was reported in two studies.52 62 Outcomes were reported at the following time points (mean±SD): preoperatively (14±14.6), 3 months (11.8±9.2), 4.5 months (5.7±4.4), 6 months (7.5±9.7), 9 months (4.1±4.7), 12 months (3.5±4.2) and 24 months (2.1±3.3). Netto et al 52 compared postoperative outcomes between arthroscopic and open Bankart repair procedures (n=50), both with postoperative EBI. Final follow-up was completed at mean time point of 37.5 months on a total of 42 study participants. Mean±SD were reported for both open (4.22±5.8) and arthroscopic (2.65±7.3) cohorts, with a significant difference between groups (p=0.031).

Pain intensity

Use of a VAS (VAS; 0=no pain; 10=maximal pain) for pain intensity was reported in four studies.37 53 54 63 Hwan and So63 reported improvements in pain intensity of 8.0 at baseline (initial, postinjury) to 2.0 at 5 months following rehabilitation with EBI alone. Additionally, Rhee and Lim53 reported an improvement in pain intensity from 2.8 at baseline (preoperative) to 1.30 at 1 year following open Bankart repair. Two studies only presented data for 13-year37 and 6-week54 follow-