Does earlier investigation of first-time traumatic shoulder dislocation lead to a reduction in the rate of recurrent dislocations?

Abstract

Background

The British Elbow and Shoulder Society (BESS) introduced national guidance for first-time traumatic shoulder instability in 2015. The aim of this case–control study was to evaluate the effect on re-dislocation rate following adoption of their imaging protocol in our department in 2016.

Method

We included patients >16 years old with first traumatic shoulder dislocations between: January 2013 to December 2013 and January 2016 to September 2016 (pre-guidance) and October 2016 to December 2019 (post-guidance). Clinical records were analysed to determine imaging and surgery rates, respective lag times and re-dislocation rates. Follow-up was set at 4 years.

Results

The study comprised 144 pre-guidance and 342 post-guidance patients. MRI arthrograms performed in < 25 s increased from 26.2% to 68.2% (p  <  0.001), with lag times reduced (p = 0.061). Ultrasound scans performed in > 40 s increased from 42.6% to 60.1% (p  <  0.05), with a significant lag time reduction (p  <  0.001). Time to surgery decreased from 432 to 249 days. Overall re-dislocations decreased from 14.6% to 8.5% (p  <  0.05), and mean dislocations from 1.33 to 1.14 (p = 0.028). In < 25 s, percentage of ≥3 dislocations decreased from 19.0% to 3.03% (p = 0.005).

Conclusions

BESS guideline implementation has resulted in increased rates of imaging and reduced lag times to imaging and surgery. Re-dislocations rates have significantly decreased.

Introduction

The shoulder is the most commonly dislocated joint in the human body due to its range of motion, shallow articular anatomy and its vulnerability during sporting trauma. ¹ Traumatic shoulder dislocations are commonly associated with either labral tears in younger patients or rotator cuff tears in older patients, increasing the likelihood of further dislocation and shoulder disability. ² Recurrent dislocations are common in younger patients usually secondary to either a soft tissue labral injury, bone loss or a combination of both.^3,4 Other studies have also confirmed increasing age to be associated with a decreased risk of recurrent dislocation.^5,6 In addition to pain and distress caused with a recurrent dislocation, there is evidence that this also leads to a higher risk of developing glenohumeral arthritis. ⁷

The British Elbow and Shoulder Society (BESS) and The British Orthopaedic Association published the traumatic anterior shoulder instability care pathway in 2015. ⁸ Before the introduction of this pathway, investigation and treatment of traumatic shoulder instability was variable and dependant on clinician experience, preference and patient factors.

Based on the 2015 BESS traumatic shoulder instability guidance, a new pathway was implemented at our unit where an urgent magnetic resonance arthrogram (MRA) was arranged for all those patients under 25 having a first-time dislocation and an ultrasound scan (USS) for those older than 40. The protocol for patients between 25 and 40 was symptom-specific, with MRA offered to those showing instability symptoms and USS to those with cuff weakness. The rationale behind early imaging was that it would facilitate early decision making and definitive management, therefore improving rates of re-dislocation. The aim of this study was to assess whether the implementation of this imaging protocol led to a reduction in the rate of re-dislocation.

Materials and methods

Patient selection

Patients over the age of 16 presenting with a first-time traumatic dislocation were identified retrospectively from a prospectively collected trauma database at a single institution for the post-guidance cohort (1 October 2016 to 31 December 2019) and retrospectively from the Emergency Department coding data between 1 of January 2013 to 31 December 2013 as well as 1 January 2016 to 31 September 2016 (Pre-guidance cohort). Follow-up was set at 4 years for each patient. Patients with previous dislocations, concurrent fractures and non-traumatic mechanisms were excluded. Figure 1 demonstrates exclusions and shows that there were 144 patients in the pre-guidance cohort and 342 in the post-guidance cohort.

Figure 1.

Flowchart demonstrating the patient selection process.

Data extraction

Electronic care records from the Emergency department, discharge summaries, radiographic review and fracture clinic letters were used to extract patient demographics such as age and gender as well as mechanism of injury, concomitant injuries, further imaging and whether they underwent surgery. Lag time from presentation to imaging and surgery and re-dislocation rates were also calculated, the latter by using information from follow-up appointments and further ED attendances.

In order to make comparisons the pre- and post-guidance cohorts were subdivided into three age categories ( < 25 s, 25–40,  >  40 years) as per the BESS first-time traumatic dislocation guidance. Pre- and post-guidance cohorts were compared both as a large group and as per age category to determine the effect of intervention.

Statistical analysis

SPSS (Version 27, IBM Corp, Armonk, NY) was used for statistical analysis. Statistical significance was set at p ≤ 0.05, with a Chi Squared or Mann–Whitney U test being used to assess this level of significance.

Our primary outcome was the effect of implementation of BESS guidance on re-dislocation rate.

Results

Demographics

Data was collected for 144 patients in the pre-guidance cohort and 342 in the post-guidance cohort. Mean age was 48.3 (SD +/- 23.3) with a range of 16–99 years. 63.2% of patients were male and 36.8% female. Distribution of patients by age is shown in Table 1.

Table 1.

Distribution of patients by age group.

Age (Years)	Pre-guidance (n/Percentage)		Post-guidance (n/Percentage)
< 25	42	29.2%	66	19.3%
25–40	34	23.6%	88	25.7%
> 40	68	47.2%	188	55.0%

Primary outcome: re-dislocations

We described re-dislocations using three different methods: mean dislocations per patient, the percentage of patients with primary dislocations that experienced re-dislocations, and < 25 s that were dislocating 3 or more times.

Overall, the percentage of patients who sustained a dislocation after the primary dislocation reduced from 14.6% to 8.5% (p  <  0.05) following the introduction of BESS guidance. Re-dislocation rates are presented by age group in Table 2.

Table 2.

Patients sustaining a re-dislocation by age group (both conservatively and surgically managed).

Age group (Years)	Pre-guidance (n/Total/Percentage)		Post-guidance (n/Total/Percentage)		p-value
< 25	10/42	23.8%	10/66	15.2%	0.259
25–40	8/34	23.5%	9/88	10.2%	0.057
> 40	3/68	4.4%	10/188	5.3%	0.771
All	21/144	14.6%	29/342	8.5%	0.043

The mean number of dislocations per person decreased from 1.33 in the pre-guidance cohort to 1.14 post-guidance (p  <  0.05) which was statistically significant. In the < 25 age group, the same measure reduced from 1.76 to 1.18 (p = 0.100) (Figure 2). The number of < 25 s sustaining ≥3 dislocations reduced from 19.0% to 3.03% (p = 0.005).

Figure 2.

Mean dislocations per person pre- and post-guidance.

The overall rate of re-dislocation in those patients managed non-operatively saw a fall from 9.32% to 4.44% (p = 0.081). This difference was most obvious in the 25–40year age group with a drop from 17.2% to 5.1% (p = 0.045).

Imaging

In the pre-guidance cohort 26.2% of < 25 s underwent an MRA with a mean lag time of 158 days (SD +/- 221.2). In the post-guidance group, the rate of MRA improved to 68.2% (p  <  0.001) with a trend in reduction of lag time to 62.1 days (SD +/- 88.8) (p = 0.061).

In terms of USS in the > 40 age group, the rate of USS increased from 42.6% to 60.1% (p  <  0.05) and the lag time decreased from mean 62.4 days (+/- 45.4) to 35.9 days (+/- 58.7) (p  <  0.001). With a greater number of patients having USS in the over 40 age group the proportion of USS showing rotator cuff tears fell from 75.9% to 53.1% (p = 0.027). The overall rate of cuff tears diagnosed in the cohort stayed the same 32.4% and 31.9% (p = 0.947) (Figures 3 and 4).

Figure 3.

Percentage of patients undergoing imaging or surgery.

Figure 4.

Mean lag time to imaging and surgery.

The 25–40year age group does not have routine imaging recommended in the BESS guidance. In the pre-guidance group, 20.6% of patients had an MRA compared to 27.3% post-guidance (p = 0.447). The rate of USS decreased from 11.8% to 4.5% (p = 0.149).

Surgery

Across all age groups, the prevalence of surgery did not change significantly with 18.1% having surgery pre-guidance and 14.3% post-guidance (p = 0.299). Lag time to surgery reduced from 423 (SD +/- 476.1) days to 250 (SD +/- 264.9) days (p = 0.072). There was no significant improvement in pre-operative re-dislocations, with 0.88 per person pre-guidance compared to 0.69 per person post-guidance (p = 0.322). Percentage of patients undergoing surgery according to age groups is summarised in Table 3.

Table 3.

Percentage of patients undergoing surgery according to age group.

Age group (Years)	Pre-guidance (n/Percentage)		Post-guidance (n/Percentage)		p-value
< 25	11	26.2%	14	21.2%	0.550
25–40	5	14.7%	10	11.4%	0.614
> 40	10	14.7%	25	13.3%	0.772
All	26	18.1%	49	14.3%	0.299

Discussion

The primary aim of this study was to determine whether the implementation of BESS shoulder instability guidance ultimately led to a reduction in re-dislocation rate. Secondary aims were to determine whether this same guidance led to an increase in the rate of imaging for those that required it and a reduction in lag time from injury to imaging and surgery. The main finding of this study was a reduction in re-dislocation rates and a reduction in those patients under 25 dislocating 3 or more times. This was likely due to the focus of our secondary aims; rate of imaging increased for the under 25 s having MRA and the over 40 s having an USS, and imaging happened with a shorter time lag, ultimately leading to earlier decision making.

The recommendation that patients over 40 have an USS led to a significant increase in number of scans being performed from 42.6% to 60.1% (p  <  0.05). There was a reduction in the proportion of scans that showed a rotator cuff tear from 75.9% to 53.1%, but the overall rate of rotator cuff tears in both the pre-guidance and post-guidance groups stayed the same, 32.4% and 31.9%. This suggests that more liberal scanning did not necessarily identify more patients with rotator cuff tears, despite it leading to a reduction in time to imaging and surgery. The associated cost increase of over-imaging suggests that there is scope for the guidelines to be amended such that there are clearer indications for scanning those over 40. As in the under 40 s, perhaps there should be greater focus on scanning only those symptomatic with cuff weakness. Whilst cost-benefit analysis is beyond the scope of this study, it is clear that this point would feature in further work on this topic. The balance here is that more-timely scanning as seen post-guidance can improve the primary outcome. Indeed when considering repair of torn rotator cuffs, Petersen, et al. ⁹ showed that across all age groups the outcomes after repair were better if done within 4 months of injury. Hantes, et al. ¹⁰ have suggested that this is likely secondary to increasing muscle atrophy over time. This supports earlier USS in patients over the age of 40 with a traumatic shoulder dislocation. Additionally, Simank, et al. ² have shown that patient satisfaction scores following rotator cuff tears due to a dislocation are better if treated surgically.

Although the reduction in time to MRA from 158 days to 62.1 was notable, this did not reach statistical significance (p = 0.061). The significant improvements seen in primary outcome in the < 25 s suggests that the increased MRA uptake (26.2% to 68.2%, p  <  0.001) was beneficial and we would therefore support early MRAs in this age group. When weighing this up it is important to consider the significant cost burden that comes with such MRA uptake as well as the limitations in its sensitivity, found to be 65% by Jonas, et al. ¹¹

Although there was a reduction in time from injury to surgery, the majority of this was seen in time to imaging and not from imaging to surgery. This suggests that the improvements we have seen in re-dislocation rates with earlier imaging could potentially be even more notable if surgical waiting times are reduced. This would further contribute to the reduction in the number of dislocations prior to surgery that we have seen between cohorts.

The rate of surgery remained the same following implementation of BESS guidelines. This suggests that the improvement in re-dislocation rates came from a reduction in time to surgery and not just from identifying new patients that were previously not being offered surgery. Indeed, the aim of this pathway at our unit was not to offer surgery to those dislocating for the first time or be more liberal with surgery. The aim was to expedite decision making through identifying those that need surgery in a more-timely fashion.

When interpreting the concepts above; time to surgery and patient selection for surgery, it is important to consider how these then lead to an improvement in our primary outcome measure. Reducing recurrent dislocations of the shoulder is likely to reduce damage caused to the shoulder joint in the longer term. Hovelius, et al. ⁷ showed that recurrent dislocations were a risk factor for developing glenohumeral joint OA on long-term follow-up. There is also evidence to suggest that reducing time between injury and surgery can reduce the risk of recurrent dislocations with Porcellini, et al. ¹² showing an interval greater than 6 months between injury and surgery being a risk factor for recurrence and Arciero, et al. ¹³ showing a reduction in recurrent instability in the < 24 s if treated with a Bankart repair following the primary dislocation. This does lead to a debate about whether first-time dislocations should be offered surgery. Currently, this is an issue under significant debate, and this study was not designed to answer this question. The current practice of our unit is not to routinely offer surgery to first-time dislocations, but rather if patients are symptomatic with physiotherapy or experience further instability. However, Dickens, et al. ¹⁴ have argued that waiting for repeat instability can lead to more glenoid bone loss and change the type of surgery required. When considering this and previous work suggesting a higher risk of arthritis with recurrent dislocations perhaps this is a time to reconsider our practice.

This has to be balanced with evidence from Hovelius, et al. ⁷ that showed that most shoulders in the < 25 age group (65%) will likely stabilise over time and a trial of conservative management is therefore a reasonable course of action with surgery for everyone under 25 likely leading to unnecessary surgeries.

Alongside the changes in surgical intervention, there was a reduction in re-dislocations in non-operatively managed patients (9.32% to 4.44%, p = 0.081), which was significant in the 25–40 year age group (17.2% to 5.1%, p = 0.045). A possible explanation for this is that the symptom-based approach in this age group of offering imaging in our department, whereby patients with instability symptoms were offered an MRI arthrogram and those with cuff weakness an USS, successfully selected appropriate patients out of surgery. The finding of Jonas, et al. ¹¹ of 100% specificity of MRI arthrograms for labral tears contributes to this idea of selecting out unnecessary patients for surgery. Whilst the 25 to 40 age group is not covered well in the BESS guidance, part of implementing these guidelines was improving information for patients, earlier treatment with physio and investigations, which together has likely helped this group of patients indirectly.

The main limitation of this study is that instability usually presents as a spectrum of symptoms and this is better captured using instability specific patient-reported outcome measures. We have only concentrated on re-dislocations as an outcome measure. In future studies, it may be useful to include patient-reported outcome measures such as Oxford Instability Score, as this would detect patients self-managing with instability symptoms but no redislocation. Another limitation of the study is the use of two cohorts separated by time in the pre-guidance group of patients. This limitation was due to availability of data but is unlikely to affect the results and conclusions as changes to management occurred in October 2016 with little difference in the two groups pre-guidance. Another limitation is the difference in group sizes as the post-guidance cohort was twice as large as the pre-guidance cohort.

There is a question of loss of follow-up. Calculating loss to follow-up in this study posed a challenge as only dislocation-positive follow-up was recorded, i.e. clinic appointments for routine follow-up were not. We have made an assumption that these patients would re-present to our centre rather than other ED departments. However, we looked through records including clinic records so dislocations in patients who presented elsewhere and were referred back to our centre would have been picked up. Our hospital like many large teaching hospitals has a captive population with some overlap with other units, but largely a captive population. Loss to follow-up is an issue with patients that recurrently dislocate as their non-attendance rate does tend to be higher. This is not likely to impact our conclusions, but does probably answer why our rate of MRA in the younger cohort was not higher after the implementation of the guidance.

The only missing data was follow-up dates for all patients. However, all fracture clinic follow-ups and shoulder X-rays were reviewed, and dislocation-positive attendances recorded. All dislocations which occurred within 4 years of the primary dislocation presentation were recorded. 4 years was chosen as it is a commonly used time frame in the available literature and suited our data collection time frame. Complete case analysis would not be suitable here as it is not possible to determine which patients are missing data. Missed follow-up in this case would be missing not at random as those without further dislocations are more likely to miss clinic follow-up.

This study poses questions that may be answered with further research. It would be interesting to see whether a reduction in waiting times to surgery on top of the reduction to imaging further improved rates of recurrence.

Additionally, greater focus on the utility of MRA arthrograms will be required to make a solid argument that their significant cost is worthwhile. There is a strong focus on their utility in assessing Bankart and SLAP lesions, however, less conviction on their predictive value of osseous lesions on instability recurrence. Whilst such lesions have been associated with a greater risk of recurrence, the ability of MRAs to appropriately assess the size and morphology of these compared to CT scans remains debated. ¹⁵

Conclusion

BESS guidance on first-time traumatic shoulder dislocation has led to reductions in mean dislocations per person, the percentage of patients with primary dislocations that experienced re-dislocations, and the number of those under the age of 25 dislocating ≥ 3 times.