This study is in response to a research commission from the National Institute for Health Research (NIHR) Health Technology Assessment (HTA) programme and, as such, this avenue of research has already been deemed necessary. Since the commissioned call, no systematic reviews or randomised clinical trials that answer the brief have been published.
The most common joint dislocations seen in hospital accident and emergency departments affect the shoulder (8.2–17 cases per 100,000 people per year).1 Around 95% of traumatic dislocations of the shoulder occur anteriorly, where the top end of the arm bone (humerus) is forced frontwards out of the shoulder socket.1 The mobility of the shoulder joint renders it particularly unstable and susceptible to re-dislocation. Traumatic anterior shoulder dislocation (TASD) is particularly common in younger patients and often occurs as a result of injury during contact sports.2 When it occurs, it is very painful and the shoulder often stays dislocated until it is repositioned. The condition is associated with significant morbidity as, following a first-time dislocation, there will probably be damage to the soft tissue and ligaments that are responsible for stabilising the joint, rendering it susceptible to re-dislocation. The literature reports that recurrent dislocation can occur in 85–92% of cases.3 However, the most effective treatment for the management of first-time traumatic shoulder dislocation in preventing further dislocations remains uncertain.
Surgery versus conservative treatment
Current options for the management of TASD include surgical or conservative treatment (usually physiotherapy) that aims to restore the stability and function of the shoulder joint.4 However, there is a lack of consensus and a lack of good-quality evidence in support of a particular treatment regime.2 Prior to both surgical and non-surgical intervention, closed reduction techniques tend to be used to restore the correct position of the shoulder joint.4 Subsequent surgical management tends to include either soft-tissue reconstruction (e.g. Bankart labral repair) or bony procedures (e.g. coracoid process transfer).5 Alternatively, non-surgical treatments involve immobilisation of the arm using slings or splints, followed by physical rehabilitation.2 It is currently unclear from the literature which treatment approach to use following a first-time TASD to restore the stability and function of the shoulder and to help prevent recurrent dislocations.
The use of traditional conservative management approaches after initial reduction and joint immobilisation has been challenged because of high rates of recurrent dislocation among some population groups. In younger patients, rates of recurrence as high as 92–96% have been reported.6 An incidence study of shoulder instability among athletes at a US military academy showed that 85% of athletes experienced a recurrent event within a 9-month period.7 A systematic review showed that there were some limited data to support primary surgery following a first-time TASD among young adults engaged in demanding physical activities (military personnel and athletes).5 A later systematic review also showed that among younger patients, a significantly lower rate of recurrent instability was identified in a 2-year period following a first-time TASD for those having surgery than for those having no surgery (7% vs. 46%).8 Consequently, there appears to be some limited evidence for surgical intervention following a first-time TASD among younger and/or highly active patients; however, the literature emphasises that there is no evidence to challenge the use of non-surgical techniques for other patient groups.5
Concerning non-operative treatment approaches in the management of first-time TASD, not only is there a lack of evidence for non-surgical over surgical treatment, but there are also uncertainties regarding the type of non-operative treatment used. For example, there is debate over the length of time the arm should be immobilised and the position (i.e. internal or external rotation) in which it should be immobilised.6 Some studies have found a lower recurrence rate in patients treated using external rotation (26% recurrence) than in those treated using internal rotation (42% recurrence) methods, and that this technique was also more effective for the younger, < 30 years age group.9 However, an earlier systematic review did not identify any statistically significant results in re-dislocation rates among patients treated using internal or external rotation methods.2 The literature has highlighted the absence of and the usefulness of future trials looking at these aspects of non-operative management for TASD.2
The use of surgical intervention for the management of TASD goes back to 1923, when Bankart described an anterior labral avulsion of the glenoid during shoulder dislocation.10 Current approaches involve stabilising the joint using open or arthroscopic (keyhole) surgery; however, the literature is unclear as to which strategy is most effective. No significant differences have been identified between open and arthroscopic approaches in terms of recurrent instability or re-injury.8,11,12
Incidence studies
Studies of the incidence of traumatic shoulder dislocation have been conducted outside the UK. An early, highly cited study of the incidence of shoulder dislocation was carried out in Sweden in 1982 by Hovelius.13 In a random sample of 2092 people aged 18–70 years, it was shown that 1.7% of participants had a history of dislocation, with re-dislocation more common in young adults and with a male-to-female ratio of 3 : 1 overall (although varying with age).13 In a 10-year follow-up study of 247 Swedish patients aged 12–22 years at the time of their dislocation, 66% of patients had one or more re-dislocations but only 24% had a recurrence between 30 and 40 years of age.14
The incidence of shoulder dislocation was again examined in a later (2010) study based on a US population.15 This study utilised data from the National Electronic Injury Surveillance System (NEISS) and was based on patients of all ages who experienced a shoulder dislocation from 2002 to 2006. Their findings showed an overall adjusted incidence rate of 23.9 per 100,000 person-years [95% confidence interval (CI) 20.8 to 27.0 per 100,000 person-years], a rate that was more than double that originally thought. The majority of dislocations occurred in men (72%), with the highest incidence observed in those aged 20–29 years (47.8 per 100,000 person-years, 95% CI 41.0 to 54.4 per 100,000 person-years). In males, the overall incidence rate was 34.9 per 100,000 person-years (95% CI 30.1 to 39.7 per 100,000 person-years), whereas in females this was 13.3 per 100,000 person-years (95% CI 11.6 to 15.0 per 100,000 person-years).
A further study, by Leroux et al.16 in 2014, also looked at the incidence rate of primary anterior shoulder dislocation in a Canadian population of 16- to 70-year-olds who were diagnosed between April 2002 and September 2010. Compared with the US study, the Canadian data showed a similar rate of dislocations in men (74%), with an incidence rate highest for 16- to 20-year-olds (98.3 per 100,000 person-years). Similar figures for the overall adjusted incidence rate were observed, which in males was 34.3 per 100,000 person-years and in females was 11.8 per 100,000 person-years.
It is unclear in the current literature as to the most effective treatment pathway (i.e. surgery vs. no surgery) in the management of first-time TASD. Regarding conservative treatment, the optimum position for arm immobilisation and the duration of time are still in question. Concerning surgery, it is debated what technique (i.e. open or arthroscopic, soft tissue or bony) is more effective, and when or if surgery is needed following a first-time TASD. The main problem is an absence of data on the natural history of shoulder dislocation, including in the UK where age and sex incidence data have not been published. The literature also highlights the lack of good-quality evidence and supports the need for further research and randomised trials to address these issues.
Aims
The commissioned aims were:
- to study the association between surgical treatment and recurrence rates following a first-time TASD in young adults who had surgery compared with those who had not had surgery
- to identify clinical predictors of recurrent dislocation in young adults with a TASD for surgical and non-surgical patients.
Objectives
To use routinely collected data from two NHS computerised databases [i.e. the Clinical Practice Research Datalink (CPRD) and Hospital Episode Statistics (HES)] to study the association between surgical treatment and rates of re-dislocation, compared with no surgery in young adults with a first-time TASD. Potential predictors of re-dislocation in patients from each treatment group were further investigated.
The research questions were addressed by implementing a two-stage approach using two work packages.
Work package 1
Work package 1 consists of an internal and external validation study to test the quality and completeness of coding in the CPRD for identifying patients aged 16–35 years diagnosed with and treated for a first-time TASD. From these data, age and sex prevalence rates for first-time TASD in the UK were produced and these were externally validated against reported rates published in other settings.
Work package 2
A propensity-score-matched cohort analysis was conducted using CPRD and HES data. The cohort of participants used in the analysis comprised young adults (aged 16–35 years) with a TASD, who had at least 2 years of coding in the CPRD prior to a first-time entry Read code for shoulder dislocation (washout period) and at least 2 years of follow-up coding. The association between treatment strategy (i.e. surgery compared with no surgery) and rates of re-dislocation were then studied. Propensity matching ensured that patients undergoing surgery were matched and compared with a non-surgical control patient. Risk factors that may play an important role in re-dislocation in both the surgical and the non-surgical groups were further investigated.
Methods
Work package 1
The first phase of this project involved conducting an internal and external validation study to test the suitability of using the CPRD data set for identifying patients with a first-time TASD, and then externally validating the findings against published results from other settings. Relevant risk factors that were identifiable in the CPRD were recorded and used to inform the formal analysis regarding future predictors.
Internal validation
An internal validation study was conducted to check the quality of the coding for shoulder dislocations and treatments in the CPRD. A cohort of patients aged 16–70 years who were diagnosed with a shoulder dislocation in the UK between 1995 and 2015 were initially identified from the CPRD, to use as UK incidence data for all age groups. The included patients all had at least 2 years of coding in the CPRD prior to a first-time entry Read code for shoulder dislocation and at least another 2 years of subsequent coding. The internal validation exercise then focused on the planned study cohort of patients identified from the CPRD who were aged 16–35 years and had the same washout period.
A general practitioner (GP) questionnaire was designed with the help of GPs to internally validate the coding of shoulder dislocations and treatments in the CPRD. A random sample of 172 patients was then selected from those identified as meeting the above selection criteria. A questionnaire was sent to the general practice of each patient using the CPRD GP questionnaire service. A clinician at the practice completed the questionnaire by comparing the records on the CPRD with the clinical records of the patient. Written reminders to complete the questionnaire were sent to the general practice by the CPRD every 2 weeks (up to a total of four reminders). The data from returned questionnaire responses were double-entered into a database, and an academic orthopaedic shoulder surgeon was consulted to resolve data input queries.
The following criteria were established a priori to ensure that shoulder dislocation coding in the CPRD was of high quality prior to moving forwards with the main analysis:
- The coding of shoulder dislocation within the CPRD needed to have a positive predictive value of ≥ 75%.
- The coding of ‘primary’ or ‘first-time’ shoulder dislocation coding in the CPRD also had to have a positive predictive value of ≥ 75%.
External validation
The external validation exercise compared the age and sex incidence rates for TASD produced for the UK with those reported in other settings. For this analysis, the original cohort of patients aged 16–70 years who were identified from the CPRD with a shoulder dislocation between 1995 and 2015 were used. The CPRD has a representative coverage of around 6.9% of the UK and includes 11.3 million patients, making it broadly generalisable in terms of age, sex and ethnicity for the UK population as a whole. The external validation study itself produced population-based age- and sex-specific incidence rates for TASD for the UK. Comparing these with the published rates reported from other settings allows for external validation of the UK data.
Work package 2
Propensity-score-matched cohort analysis
The main study is a population-based propensity-score-matched cohort study comparing the association between surgery (vs. no surgery) and rates of re-dislocation in patients diagnosed with a TASD. The cohort of patients used for this analysis consisted of young adults aged 16–35 years with a TASD, with 2 years of coding in the CPRD prior to a first-time entry Read code for shoulder dislocation, and at least another 2 years of follow-up coding after the initial event. A pre-agreed list of Read codes (CPRD) and HES Office of Population Censuses and Surveys (OPCS) 4.7 codes for shoulder dislocation and treatments was used to collect all related outcomes and events; these were further informed by the earlier validation work (work package 1) (see Appendix 1). A ‘primary’ or ‘first-time’ TASD is defined here as a first-time entry Read code for shoulder dislocation.
Identified patients were allocated to the intervention (surgical) or control (non-surgical) groups. Patients in the intervention group were those who underwent shoulder stabilisation surgery following a first-time episode of TASD (early surgical repair is defined here as a ‘decision to treat surgically after a first-time TASD’). Patients in the control group were those who did not receive a surgical intervention following a first-time episode of TASD.
Propensity score matching methods were used to match each patient receiving surgery to a comparable patient in the non-surgical group. Propensity score methods were used because they provide the best approach to handling observational data sets that may be influenced by confounding (e.g. some patients being more likely to have surgery than others). Propensity score methods allow for bias being introduced into the data set through confounding, as the type of treatment received (i.e. surgery or no surgery) was not randomly allocated in this study.
After the process of matching surgical patients to non-surgical control patients, a Cox regression survival model was used to assess the association between surgery and time to re-dislocation over a 2-year period.
Identify clinical predictors of recurrent dislocations by treatment type
In this component, investigated potential risk factors associated with re-dislocation for both the surgical and the non-surgical groups were investigated. Prediction models were developed using linked CPRD and HES data, including any risk factors defined a priori by consensus that were available and those identified through the earlier validation work.
Conclusion
The relevant background information supporting the need for research into the efficacy of management options for patients with first-time traumatic shoulder dislocation has been described. Each of the four key components of the study has been outlined in this chapter and they are described in more detail in Chapters 2–5.