High Variability of the Definition of Recurrent Glenohumeral Instability: An Analysis of the Current Literature by a Systematic Review

Purpose To determine the definitions for recurrence used in the literature, assess the consensus in using these definitions, and determine the impact of these definitions on recurrence rates. Methods A literature search was performed in PubMed and EMBASE including studies from 2000 to 2020 reporting on recurrence rates after anterior arthroscopic shoulder instability surgery. Dislocation, apprehension, subluxation and recurrence rates were compared. Results Ninety-one studies were included. In 68% of the eligible studies, recurrence rates are not well defined. Thirty (33%) studies did not report on dislocations, 45 (49%) did not report on subluxations, and 58 (64%) did not report on apprehension. Seventeen different definitions for recurrence of instability, 4 definitions of dislocations, and 8 definitions of subluxation were used. Conclusion Recurrence rates are poorly specified and likely underreported in the literature, hampering comparison with results of other studies. This highlights the need for a consensus on definition of recurrence across shoulder instability studies. We recommend not using the definition recurrence of instability anymore. We endorse defining dislocations as a radiographically confirmed dislocation or a dislocation that is manually reduced, subluxations as the feeling of a dislocation that can be (spontaneously) reduced without the need for a radiographically confirmed dislocation, and a positive apprehension sign as fear of imminent dislocation when placing the arm in abduction and external rotation during physical examination. Reporting on the events resulting in a dislocation or subluxation aids in making an estimation of the severity of instability. Level of Evidence Level IV, systematic review.

D epending on the risks for recurrent shoulder instability can be managed conservatively, with (arthroscopic) soft-tissue procedure, or (open) bony procedures. 1 The arthroscopic Bankart repair is the most used procedure, including up to 87% of instability procedures. 2 Several studies have assessed recurrence rates after shoulder instability surgery. The recurrence rate for the general population varies from 0% to 8% after Latarjet procedure 3 to 3.4% to 35% after arthroscopic Bankart repair. 4,5 Although most studies describe rates of recurrent dislocation (instability), there is no consensus on the definition of these terms. For example, Randelli et al. 5 uses redislocation or subluxation as a definition of recurrent instability, whereas Gerometta et al 4 does not describe a definition of a recurrence of instability/dislocation. As a result, findings in previous studies were hard to compare. 6 Kuhn 7 has described that shoulder instability studies are procedure based and not condition based, resulting in unclear definitions of instability. He introduced the frequency, etiology, direction, and severity system for describing instability. Kennedy et al. 8 has described that there is a wide variety of definitions of recurrence in the literature and that the recurrence rates vary according to level of evidence, age, follow-up time, and attrition rate. Although Kennedy et al. 8 have noticed that there are many different definitions used in the literature, it is still unclear how many studies did not define these definitions. The purposes of our study were to determine the definitions for recurrence used in the literature, assess the consensus in using these definitions, and determine the impact of these definitions on recurrence rates. We hypothesized that for shoulder instability the definition of recurrence is poorly reported and that there is no consensus on the definition to be used.

Methods
This is a systematic review of available literature on the definition of recurrent anterior shoulder instability and is performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline. 9 No review protocol was identified for this study.

Literature Search and Study Selection
A literature search was performed on August 5, 2020, in PubMed and EMBASE with predefined search terms (Appendix 1), including all studies mentioning recurrence rates in Dutch, German, Arabic, and English. The search was limited to studies between 2000 and 2020 to give insight into definition of recurrence in the most recent literature. The inclusion criteria included studies assessing recurrence rates after arthroscopic anterior shoulder instability surgery. Letters to the editors, instructional courses, animal/cadaver/in vitro studies,  99 62 44 24 76 72 77 45 78 36

Data Extraction
First, we checked whether the authors reported on recurrence rates and how they defined recurrence of instability, subluxations, and dislocations. Afterward, the recurrence rates, dislocation rate, subluxation rate, and positive apprehension rate were extracted and presented. The methodological quality of each study was assessed separately by the same 2 authors using the Coleman Methodology Score. 10 The total number of points correlates with poor (0-49 points), fair (50-69 points), good (70-84 points), or excellent (85-100 points) quality of the study.

Study Selection
In total, 2,569 titles and abstracts were screened, from which 383 studies were full-text screened resulting in 89 studies being included in the final analysis (Fig 1). From the 282 studies that were eligible for inclusion, 193 (68%) were excluded because the definition of recurrence was not defined clearly. Cross-referencing resulted in inclusion of 2 additional studies. The 2 observers agreed on 83.7% of the articles with a Cohen's kappa of 0.67.

Recurrence Rates Reported
Recurrence rates, as well as the rates of dislocation, subluxation and positive apprehension test results for each article, are reported in Table 2. In Park et al., 56 we could not extract the exact number of recurrent dislocations because only the amount of dislocations in group 1 were reported. Thirty studies (33%) did not report on recurrent dislocations, 45 studies (49%) did not report on recurrent subluxation, and 58 studies (64%) did not report on apprehension.

Discussion
The results of this study show that there are no uniform definitions of recurrence, subluxation, or dislocation after shoulder stabilization surgery used in the current literature. Using different definitions leads to a high level of heterogeneity. This could lead to misinterpretation of results and conclusions.
y In a table the number is expressed as dislocations, while in the text as dislocations and subluxations.  24,37,44,60,69,71,99 Sense of dislocation with a positive anterior apprehension test 59 Transient instability event that did not require reduction but demonstrated a positive apprehension and relocation sign with radiographic or magnetic resonance imaging evidence of a Bankart or Hill-Sachs 19 Symptomatic self-reported subluxation 62 "Dead-arm" phenomenon or instability which spontaneously reduced 73     with signs of a sustained dislocation in further radiographs, such as Hill-Sachs or bony Bankart lesion in comparison with the preoperative situation, could be categorized as a confirmed dislocation. For subluxations, we advise using the definition of the feeling of a dislocation that can be (spontaneously) reduced without the need for a radiographically confirmed dislocation. For a positive apprehension sign, we suggest using the definition as mentioned by Lädermann et al. 100 as fear of imminent dislocation when placing the arm in abduction and external rotation during physical examination. We suggest not using the definition recurrence of instability anymore to avoid using multiple meanings of this term; if used we suggest using the definition as a dislocation or a subluxation and also report on these events separately. We chose this definition because of the fact that dislocations and subluxations can be regarded as a (partial) failure of the operation, whereas a positive apprehension test result does not always correlate with instability of the shoulder. This is because a positive apprehension could be related to changes in functional cerebral networks induced by prior instability that can persist even after stabilizing the shoulder. 101 Finally, we endorse reporting on the events resulting in a dislocation or subluxation to be able to make an estimation of the severity of instability. For example, a shoulder that dislocates during normal daily activities is potentially more unstable in comparison with a shoulder dislocating after a collision during sports.

Limitations
Although DSA and DS have significantly different recurrence rates, the high number of studies not reporting dislocations, subluxations, and apprehension rates separately makes it unknown whether the recurrence rates would remain similar if all studies held the same criteria for defining recurrences in their cohort (e.g., not including apprehension in the definition could lead in less-reported recurrences). Because of corrections for multiple comparisons being not feasible for 17 definitions and because of the high variability in surgical techniques and patient characteristics, we did not compare the results for the different definitions. Remarkably, we had to exclude 68% of eligible studies because recurrence rates were not defined at all.
Another limitation of this study is that we could not compare the results of the different techniques to assess whether other definitions could lead to other results. We agree with the results Kuhn 7 and Kennedy et al. 8 The difference with Kennedy et al. 8 is that we were stricter in whether a definition is explicitly defined to  (20) Recurrent dislocation, symptomatic subluxation or instability preventing return to full active duties or necessitating an additional surgical stabilization procedure.
119 6 (5) 9 (8) Recurrence of subluxation or frank dislocation or an ongoing or new feeling of instability 141 12 (9) 19 (13) Subluxation, 1 or more frank dislocations, or at least 1 episode of dead arm syndrome 50 3 (6) 2 (4) 5 (10) 5 (10) Dislocation or revision 71 17 (24) Percentages are percentages of total patients in studies mentioned to have an event (e.g., if a study does not report on subluxations, it is not used to calculate the percentage of subluxations) avoid overestimation of the reporting of recurrence rates; for example, in Kennedy et al. 8 an article was regarded as defining recurrences as a dislocation if they only reported on dislocations without explicitly defining recurrences. Kasik and Saper 101 have also reported that there are different definitions of recurrences after arthroscopic Bankart repair in the adolescent athletes. However, just like the article of Kennedy et al., 8 they also included articles that do not define recurrences explicitly.

Conclusion
Recurrence rates are poorly specified and likely underreported in the literature, hampering comparison with results of other studies. This highlights the need for a consensus on definition of recurrence across shoulder instability studies. We recommend not using the definition recurrence of instability anymore. We endorse defining dislocations as a radiographically confirmed dislocation or a dislocation that is manually reduced, subluxations as the feeling of a dislocation that can be (spontaneously) reduced without the need for a radiographically confirmed dislocation, and a positive apprehension sign as fear of imminent dislocation when placing the arm in abduction and external rotation during physical examination. Reporting on the events resulting in a dislocation or subluxation aids in making an estimation of the severity of instability.