Abstract
Background:
Direct comparisons of the demographic and clinical risk factors between patients with anterior and posterior glenohumeral instability are uncommon.
Purpose:
To identify and compare demographic, clinical, and perioperative variables in patients receiving arthroscopic labral repair for anterior and posterior shoulder instability.
Study Design:
Case series; Level of evidence, 4.
Methods:
A retrospective chart review was performed for patients who underwent primary arthroscopy for shoulder instability by 7 surgeons at a single institution between 2012 and 2020, excluding revision surgeries and multidirectional instability patients. Demographics, radiological findings, and intraoperative data were collected. Patients with anterior instability (AI) were compared to those with posterior instability (PI) by number of dislocation events (0, 1, 2, or >2), chief complaint (dislocation event and direction, subluxation, or pain), and concomitant intraoperative procedures. A subgroup analysis was performed of patients with documented dislocations. Statistical analysis included the Student t tests and Mann-Whitney U test for continuous variables and chi-square or Fisher exact tests for discrete variables with significance defined as a P value <.05. Bonferroni corrections were applied.
Results:
A total of 482 shoulders met the inclusion criteria. Overall, 80% (384/482) of the patients were evaluated with AI and 20% (98/482) with PI. The PI group demonstrated a greater mean BMI compared with the AI group (28 ± 7 vs 26 ± 6; P = .003). There were no significant differences in age, sex, or contact/collision athlete status. Overall, 80% (308/384) of the patients with AI sustained a dislocation compared to 43% (42/98) of those with PI. A higher proportion of patients with PI (without instability) reported more pain than patients with AI (without instability) (42% vs 12 %; P < .001). Recurrent dislocations (>2) were more common in the AI group compared with the PI group (51% vs 21%; P < .001). Patients with AI underwent concomitant posterior labral repair (17% [67/384]) at a similar rate to patients with PI who underwent concomitant anterior labral repair (16% [16/98]). Subgroup analysis of patients with discrete dislocations demonstrated similar rates of those receiving concomitant posterior labral repair in the AI group when compared with those in the PI group receiving concomitant anterior labral repair (14% vs 17%).
Conclusion:
Patients arthroscopically treated for AI undergo concomitant posterior labral repair at rates similar to those with PI requiring concomitant anterior labral repair. This finding suggests that tear extension occurs at similar rates in patients with AI and those with PI. Additionally, patients with AI requiring labral repair are more likely to experience multiple dislocation events as opposed to patients with PI who present with pain and subluxation.
Keywords
Glenohumeral instability is characterized by disruption of the static and/or dynamic stabilizers of the shoulder, resulting in a spectrum of presentations from pain to transient subluxation, to dislocation of the humeral head from the glenoid fossa. The vector of this derangement can vary, occurring anteriorly, posteriorly, inferiorly, or in multiple directions.3,9,10 Presenting with symptoms and findings of glenohumeral instability typically contributes to the diagnosis and management strategy; however, occasionally the direction of instability fails to correlate with specific injury patterns.
Patients with anterior instability (AI) typically present after an anterior dislocation. 12 Risk factors for AI have been extensively studied and include young age, male sex, contact athletes, overhead athletes, ligamentous laxity, and the presence of humeral or glenoid bone loss.15,16,18 Posterior instability (PI) can be the result of dislocation or, alternatively, overuse or repetitive microtrauma without ever having experienced a dislocation.6,22 Risk factors for PI have been less commonly reported, but commonly include being in the military, being a football player, and having a history of seizure disorder.2,7 Previous studies have described the presence of concomitant anterior labral tears in patients with PI. 20 However, there remains a paucity in the current literature as to the reporting of the incidence of anterior labral repair in PI and, conversely, the incidence of posterior labral repair in AI.
The objective of this study was to identify and compare the demographic, clinical, and operative characteristics of patients receiving arthroscopic labral repair for AI and PI, with a focus on the severity and number of instability events experienced. We hypothesized that although shoulders with AI would present with higher rates of dislocation, they would require posterior labrum repair at a similar rate to shoulders with PI requiring anterior labrum repair.
Methods
A retrospective review was performed at a single academic institution after institutional review board approval. All patients who underwent primary arthroscopy for shoulder instability between 2012 and 2020 were included. Surgeries were performed by 7 orthopaedic surgeons with fellowship training in shoulder and elbow surgery or sports medicine. Patients were excluded for revision surgery, multidirectional instability, or missing/inadequate documentation. Data were collected via review of the institution's electronic medical record, Epic (Epic Systems Corporation).
Patients were classified by direction of instability (anterior, posterior, or multidirectional), determined from the chief complaint and/or clinical evaluation preoperatively. The direction of instability was then confirmed intraoperatively. Patients with AI were classified as those with either a reported history of anterior dislocation or demonstrated unidirectional findings of AI on examination (eg, positive anterior apprehension/relocation). Patients with PI were defined as those with either a reported history of posterior dislocation or demonstrated unidirectional findings of PI on examination (eg, positive Kim or jerk test). Patients with subluxation were those who did not report a history of discrete dislocation but rather a general sense of instability in a given direction, with confirmatory examination findings and imaging. Patients with multidirectional instability were defined as those who reported a history of dislocation in >1 plane or demonstrated multidirectional instability on examination. Patients were similarly defined based on their initial presentation/chief complaint as having dislocation, subluxation, or pain only, with patients with only pain having no history of dislocation or subluxation, but positive unidirectional findings on examination or imaging. All imaging was interpreted by an attending radiologist at the home institution and reviewed by the senior author (J.J.E.). Preoperative variables collected included contact/collision sports participation, tobacco use, chief complaint, number of glenohumeral dislocations (0, 1, 2, or >2), and time from first dislocation to surgery. Intraoperative data were collected from the operative report of the attending surgeon. This included location of labral repair (anterior, posterior, and/or superior labrum), number of anchors required for labral repair, and concomitant procedures performed at the discretion of the attending surgeon (remplissage, capsular plication, rotator interval closure, biceps tenodesis/tenotomy, distal clavicle excision, subacromial decompression, rotator cuff repair, and humeral avulsion of the glenohumeral ligament [HAGL] repair).
Direct comparisons for all collected variables were made between patients with AI and those with PI. A subgroup analysis was conducted for patients with a dislocation event only (patients with subluxation and pain only were excluded). Analyses were also performed stratifying by the number of dislocations (0, 1, 2, or >2).
Statistical Methods
All analyses were conducted in Excel Version 16 (Microsoft) and SPSS Version 29 (IBM Corp). Data were analyzed for normality using the Kolmogorov-Smirnov test, and parametric and nonparametric tests were used as appropriate, depending on data normality. Categorical data were compared between groups using chi-square tests and Fisher exact tests depending on cell populations. Continuous data were compared between groups using Student t tests and Mann-Whitney U tests as appropriate, depending on data normality. Two-tailed tests were performed in all cases. P values <.05 were considered significant. To account for multiple comparisons, a Bonferroni correction was applied as indicated.
Results
Patient Characteristics
A total of 633 shoulders underwent arthroscopic labral repair between January 2012 and December 2020. Of these, 482 patients met inclusion criteria as 151 patients were excluded from the study for revision surgery (n = 93), multidirectional instability (n = 44), or missing/inadequate documentation (n = 14). The final study population consisted of 368 men and 114 women with a mean age of 27 ± 10 years (range, 14-71 years). Of this cohort, 80% (384/482) had AI and 20% (98/482) had PI. No significant differences were observed in age, sex, laterality, Charlson Comorbidity Index (CCI), 21 contact/collision sports, or tobacco use between the anterior and posterior cohorts (Table 1). Patients with PI had a higher body mass index (BMI) (28 ± 7 kg/m2) compared with the patients with AI (26 ± 6 kg/m2) (P = .003) (Table 1).
Demographic Summary of Entire Cohort and Those With a History of a Discrete Dislocation a
Values are presented as percentage (n/N) or mean ± SD. The Bonferroni significance level is .004. Bold P value indicates statistical significance. AI, anterior instability; BMI, body mass index; CCI, Charlson Comorbidity Index; PI, posterior instability.
Instability History
Overall, there was more heterogeneity in the presentation of patients with PI; 43% of patients with PI had experienced a dislocation, 42% had pain only, and 15% presented with complaints of a subluxation. Comparatively, 80% of patients with AI had dislocations, 12% had pain only, and 8% had subluxation. The distributions of patients with pain, subluxation, 1 dislocation, 2 dislocations, and >2 dislocations was different between patients with AI and those with PI (P < .0001). Presentation with isolated shoulder pain was approximately 4 times higher in the PI cohort (41/98 [42%]) than in the AI cohort (46/384 [12%]) (P < .001). Dislocation was more common in the AI cohort (308/384 [80%]) than the PI cohort (42/98 [43%]) (P < .001) (Table 2). Rates of presentation with subluxation were increased in the PI cohort (15/98 [15%]) when compared with the AI cohort (308/384 [80%]); however, this finding did not reach statistical significance (Table 2). The patients with AI (195/384 [51%]) were more likely to experience recurrent dislocations (>2 dislocation events) than the PI cohort (21/98 [21%]) (P < .001).
Instability History and Chief Complaint Comparisons a
Values are presented as percentage (n/N) or mean ± SD. Bold P values indicate statistical significance. The Bonferroni significance level is .003. AI, anterior instability; HAGL, humeral avulsion of the glenohumeral ligament; NA, not applicable; PI, posterior instability.
Operative Characteristics
Repair of the glenoid labrum performed for AI utilized more anchors (4 ± 1) compared with PI (3 ± 1) (P < .001). The 360° labral (circumferential) repair was similarly performed in AI (9/384 [2%]) and PI cohorts (3/98 [3%]) (P = .716). The AI cohort required concomitant posterior labral repair in 17% (67/384) of cases. In contrast, of the patients who underwent repair for PI, 16% (16/98) required concomitant anterior labral repair. This difference was not significant (P = .860) using the Bonferroni corrected significance level of .002.
Distal clavicle excision was more common in patients with PI (3/384 [1%] anterior and 8/98 [8%] posterior) (P < .001) (Table 3). Expectedly, patients with AI were nearly 7 times more likely (14% vs 2%) to require remplissage than patients with PI (P < .001). Patients with PI were more likely to receive capsular plication than patients with AI (56% vs 39%; P = .003), although this difference between groups was not found to be significant after the Bonferroni correction. There were no significant differences in other concomitant procedures performed including HAGL repair, rotator interval closure, subacromial decompression, rotator cuff repair, or biceps tenodesis/tenotomy (Table 3).
Summary and Comparison of Intraoperative Variables and Concomitant Procedures a
Values are presented as percentage (n/N) or mean ± SD. Bold P values indicate statistical significance. The Bonferroni significance level is .002. AI, anterior instability; HAGL, humeral avulsion of the glenohumeral ligament; NA, not applicable; PI, posterior instability.
Subgroup Analysis
A subgroup analysis was performed of patients presenting with patient-reported dislocation event(s). In total, 132 patients were excluded who did not experience a dislocation. Patients with anterior dislocation comprised 88% (308/350) of the total cohort, while patients with posterior dislocation comprised 12% (42/350) of the cohort. There were no significant differences in age, sex, laterality, CCI, BMI, participation in contact/collision sports, or tobacco use between the 2 groups (Table 1). Of the patients with AI, 21% (66/308) experienced 1 dislocation, 15% (47/308) experienced 2 dislocations, and 63% (195/308) experienced >2 dislocations (Table 2). Of the patients with PI, 45% (19/42) experienced 1 dislocation, 5% (2/42) had 2 dislocations, and 50% (21/42) had >2 dislocations (Table 2). Shoulders with posterior dislocation were more likely to experience a single prior dislocation event compared with those with anterior dislocation (45% vs 21%; P < .001). Between the 2 groups, there was no difference in rates of 2 prior dislocations or >2 prior dislocations (Table 2). Time from first dislocation to surgery was 42 ± 65 months for patients with AI and 58 ± 98 months for patients with PI (P = .309) (Table 2).
Anterior labral repair was performed in 99% (304/308) of patients with anterior dislocation, while posterior labral repair was performed in 100% (42/42) of patients with posterior dislocation (Table 3). Of patients with posterior dislocation undergoing arthroscopy, 17% (7/42) required concomitant anterior labral repair. In contrast, of patients with anterior dislocation undergoing labral repair, 14% (43/308) required concomitant posterior labral repair in addition to anterior repair. All patients with reported anterior dislocation who underwent posterior repair (43/43) underwent concomitant anterior repair. The 360° (circumferential) repair was not observed to be different between the 2 groups, with 2% (7/308) of patients in the anterior group and 5% (2/42) patients in the posterior group (P = .346).
Discussion
This study of 482 patients provides an extensive comparison of anterior or posterior glenohumeral instability populations. Our hypothesis was confirmed and is consistent with previous literature suggesting that patients with AI commonly present with dislocation, while patients with PI present heterogeneously with pain and instability events. Patients with PI required similar rates of concomitant anterior labral repair to those with AI requiring concomitant posterior labral repair. This holds true even when considering only patients with a history of frank dislocation. Overall, our study offers demographic and clinical insights for preoperative planning and enhances understanding of the pathogenesis of shoulder instability.
The existing literature directly comparing AI and PI has been limited to smaller cohorts, including 2 notable studies by Bernhardson et al 4 and Teske et al, 26 totaling 200 patients. Our data align with findings from Bernhardson et al and Teske et al that patients with AI tend to present with dislocations, while patients with PI commonly present with pain.4,26 Our findings, however, indicate more heterogeneity in the presentation of patients with PI, with high rates of both dislocation (43%) and pain (42%) in this cohort. The aforementioned studies reported dislocations in only 4% to 10% of patients with PI.4,26 This discrepancy may be attributed to differences in patient population. Studies by Bernhardson et al and Teske et al evaluated athletes and military recruits, whereas our study was conducted at an academic tertiary referral center. It is possible that our data capture a higher proportion of patients with traumatic dislocation referred to our center. A further finding in our data was that the proportion of patients with PI (20%) is similar to that in recent reports (24%) in high-risk populations.16,24 This is higher than the traditionally reported 2% to 12% in the literature.1,14,25 It is important to note, however, that this cohort included only patients who were surgically treated, and thus the true incidence of all patients with PI may be different.
The data also demonstrated that patients with AI are more likely to experience multiple dislocations. It is well documented in the literature that patients with AI are more likely to experience recurrent dislocations, which aligns with our data showing that 51% of patients with AI had recurrent (>2) dislocations compared with 21% of patients with PI.13,23 However, data on rate of posterior dislocations are scarce. In our sample, patients with PI had 1 dislocation 45% of the time compared with 21% of patients with anterior dislocation. When discussing PI rates, many studies include those who experience subluxation, those with pain, and those who experience a traumatic dislocation.5,13,17 Therefore these are data valuable in establishing rates of single versus multiple dislocations in arthroscopically treated patients with PI.
The mean BMI of the PI group was significantly higher than that of the AI group. While no other demographic variables differed significantly between these cohorts, trends in contact/collision athlete status and BMI may still provide useful clinical insights. Previous literature indicates that participation in contact/collision sports is a risk factor for both AI and PI.5,6,8,27 Based on this, we anticipated a higher proportion of contact/collision athletes in the PI cohort—particularly given that posterior labral tears after repeated posterior loading are well documented in football players. 28 However, our results showed similar rates of contact/collision athletes in both cohorts. Notably, the PI cohort demonstrated a higher mean BMI overall. Although BMI does not serve as an independent predictor of shoulder instability, our data suggest that higher BMI may be characteristic of patient populations at risk for repeated posterior loading, such as football players (eg, offensive linemen with repeated anterior collisions on outstretched arms), weight lifters, and military personnel.15,17
The analysis of concomitant procedures revealed a rate of 16% to 17% required concomitant “additional” labral repair for both patients with AI and those with PI. In particular, surgeons elected to repair the posterior labrum in those with AI and the anterior labrum in those with PI at a rate of 1 in every 6 patients. The subgroup analysis of patients with documented dislocation was consistent with this finding, as patients with posterior dislocations had concomitant anterior repair 17% of the time and patients with anterior dislocations had concomitant posterior repair in 14% of patients. One previous study reported that 19% of patients undergoing arthroscopic Bankart stabilization required additional posterior labral repair, which is comparable to our rate of 17%. 11 However, we are unaware of any studies that have reported the rates of concomitant anterior repair in patients with PI/dislocation. While both directions of dislocation are typically associated with traumatic mechanisms, the current literature does not establish a correlation between the extent of labral damage and the direction of dislocation. 15 These findings suggest that anterior and posterior dislocation events may exert comparable levels of trauma on the labrum,19,24 and therefore surgeons should be aware that concomitant repair of the opposite side of the labrum may be necessary in all cases, regardless of the clinically determined direction of instability.
Limitations
This study is subject to several limitations. First, its retrospective nature may pose challenges in defining the direction of instability. Efforts to mitigate this limitation were made by cross-referencing information (chief complaint, history, examination, and radiographic interpretation) from multiple locations within the medical records to establish a consensus on the direction of instability. Additionally, reliance on the history and physical examination may not capture all instances of dislocation or instability; however, similar efforts were made in this regard to avoid misinterpretation of classification. Another potential limitation arises from the fact that surgical management was carried out by various surgeons, introducing the possibility of surgeon-specific bias in the selection of treatments. Furthermore, this data set does not include the proportion of patients with shoulder instability who underwent nonoperative treatment and therefore could introduce selection bias. Lastly, our population included trauma patients, high school athletes, recreational athletes, sedentary individuals, and collegiate and professional athletes and therefore represents a mixed group of patients. Additionally, patients of older age with massive rotator cuff tears causing dislocation were not excluded and may introduce some bias in regard to this population subset. However, we feel that the larger size of this study with a younger mean age of those included mitigates any potential bias that this may introduce. Therefore, while these data capture a global population, they may not be readily translatable to high-risk populations that are traditionally considered with shoulder instability.
Conclusion
Patients arthroscopically treated for AI undergo concomitant posterior labral repair at rates similar to those with PI requiring concomitant anterior labral repair. This finding suggests that tear extension occurs at similar rates in patients with AI and those with PI. Additionally, patients with AI requiring labral repair are more likely to experience multiple dislocation events compared to patients with PI, who more often present with pain and subluxation.
Footnotes
Final revision submitted January 27, 2025; accepted February 17, 2025.
One or more of the authors has declared the following potential conflict of interest or source of funding: P.N.C. has received consulting fees from DePuy-Mitek, Exactech, DJ Orthopaedics, and Smith & Nephew; intellectual property royalties from DePuy, Exactech, and Responsive Arthroscopy; and publishing royalties from the Journal of Shoulder and Elbow Surgery; he holds stock in TitinKM Biomedical. J.J.E. has received consulting fees from Johnson & Johnson DePuy Mitek Sports Medicine, and he is an editorial board member for Arthroscopy. AOSSM checks author disclosures against the Open Payments Database (OPD). AOSSM has not conducted an independent investigation on the OPD and disclaims any liability or responsibility relating thereto
Ethical approval for this study was waived by the University of Utah (IRB No. 00169833).