Risk factors for shoulder re-dislocation after arthroscopic Bankart repair

We treated 102 consecutive shoulders (100 patients) using ABR for traumatic anterior shoulder instability from February 2002 to December 2010 at our institute. The average patient age and follow-up period was 25.7 (range, 14–40) years and 67.5 (range, 24.5–120) months, respectively. Inclusion criteria included: (1) recurrent anterior shoulder instability after an apparent traumatic episode, (2) at least three dislocation/subluxations before the surgery, (3) a Bankart lesion or anterior labral periosteal sleeve avulsion lesion confirmed during arthroscopy, (4) an arthroscopic capsulolabral repair achieved using three or more suture anchors, and (5) a minimum of 2 years follow-up was completed by telephone interview. Exclusion criteria included: (1) multidirectional instability, (2) revision Bankart repairs, and (3) full-thickness rotator cuff tears.

Preoperative radiographic imaging, consisting of anteroposterior, scapular Y, and axillary views, was obtained to evaluate the glenoid shape of the glenoid and the presence of any bony (i.e., Bankart or Hill-Sachs) lesions. Contrast magnetic resonance imaging of the affected shoulder was evaluated for the presence of a Bankart lesion and any other shoulder injury before surgery.

Institutional review board approval by the Kurume University Ethic Committee and verbal consent at the time of phone interview were obtained.

A single surgeon (MG) performed all the ABR procedures, using the same procedure, with the patients under general anesthesia in a beach chair position. A standard posterior portal was created and used as a viewing portal. Subsequently, the glenohumeral joint was inspected and pathology verified. Anterior and anterosuperior portals were established above the subscapularis tendon and anterolateral to the acromion, respectively. After mobilization of the inferior glenohumeral ligament complex from the glenoid neck up to a 7 to 5 o'clock position, the glenoid neck and its articular edge were decorticated with a motorized shaver and ring curette to facilitate healing of the repaired capsulolabrum.

The first anchor was placed at the 5 o'clock position (right shoulder), using a biodegradable push-in suture anchor (Panalock, 2002–2005 and Panalock loop, approximately 2005, DePuy Mitek, Inc., Raynham, MA, USA) with an eyelet single loaded with a no. 2 non-absorbable suture (2002–2010, Ethibond, DePuy Mitek, Inc.; 2010–2012: FiberWire, Arthlex, Inc., Naples, FL, USA). To pass the suture through the anterior portal, a shuttle relay was passed through the labrum and brought out through the anterosuperior portal. The capsulolabrum was shifted as superior to the glenoid as possible. A standard sliding knot was tied on the capsulolabral side. These steps were repeated for each anchor used in the repair and capsular shift. Three to six anchors were used for the capsulolabral repair.

The patients were kept in a shoulder sling with an abduction pillow at neutral rotation and 20° abduction postoperatively. Three weeks after the surgery, progressive self-assisted shoulder elevation and external rotation were initiated. Active range-of-motion exercise was permitted at 6 weeks postoperatively, rotator cuff strengthening at 12 weeks postoperatively, and full participation in sports at 6 months postoperatively.

We successfully contacted all 100 patients who underwent ABR in our institution via telephone. A re-visit for postoperative evaluation was requested although most of the patients declined the visit. Therefore, the patients' present status, including postoperative injury and re-dislocation with either subluxation or complete dislocation, was inquired via phone. The mean follow-up time of the phone survey was at 67.5 months (range, 24.5–120 months).

After collecting the patient's information, correlations of several risk factors were determined, including gender, injured side, age at first dislocation and surgery, arm dominance, type of sport (collision, contact, overhead, or others), waiting time prior to surgery, number of dislocations preoperatively, number of suture anchors used, superior labrum anterior and posterior (SLAP) lesion, and tear of the capsular. The patient demographic data are shown in Table 1.

Since previous studies have suggested that both glenoidal and humeral head bone defects are closely associated with re-dislocation after ABR [[6],[12],[13]], these bony defects were measured using an arthroscopic probe technique [[5]]. For glenoid bone defects, using the anterosuperior-viewing portal, a probe with 3-mm calibrated marks was placed through the posterior portal across the glenoid so that its tip rested on the bare spot. The distance from the center of the bare spot to the posterior glenoid rim was then measured. The probe was then used to measure the distance from the anterior glenoid rim to the center of the bare spot. Finally, the probe was used to measure the distance from the center of the bare spot to the inferior glenoid rim [[14],[15]]. Humeral head defects (Hill-Sachs lesions) were also measured with arthroscopic probe techniques, based on an estimation of the width, depth, and length, as measured intraoperatively with the arthroscopic probe [[5]]. The critical size of a Hill-Sachs lesion that causes instability is thought to be a volume > 250 mm³ [[16]–[18]]; thus, such lesions described ‘large Hill-Sachs lesions’.

The software JMP (SAS Institute Inc., Cary, NC, USA) was used for statistical analysis. Since the development of re-dislocation was a time-dependent outcome variable, we used survival methodology (Kaplan-Meier curve) to examine the probability of re-dislocation occurring after ABR, by setting re-dislocation as the end-point. Student's t test or chi-squared test was used to compare the bony defect size between the patients with or without re-dislocation. A chi-squared test was used to examine the correlations between the clinical parameters and re-dislocation after ABR. Logistic multivariate analysis was then performed to further evaluate the significant parameters obtained from the Pearson's chi-squared test, accompanied by the odds ratio with 95% confidence intervals. The data are expressed as the mean values with the standard deviation. A P value < 0.05 was considered significant.

Of the 102 shoulders treated with ABR, a total of 9 (8.8%) experienced re-dislocation (Figure 1). Of these, seven shoulders were re-injured within the first year with the arm elevated at 90° and externally rotated at 90°. Another experienced re-injury and re-dislocation at 15 months and 6 years after surgery. Thus, most re-dislocations (78%) occurred within the first year after ABR. Of the nine patients who had a re-dislocation, two patients underwent re-operation, and the remaining seven patients were treated non-operatively or refused operation. Of the 93 shoulders without re-dislocation, 7 shoulders had a traumatic injury within the first year under the same conditions (90° elevation and 90° external rotation). The shoulders were re-dislocated during contact and overhead sports (n = 2), as well as activities of daily livings (n = 5). Thus, re-injury within the first year proved to be a risk for re-dislocation after ABR (P < 0.001, chi-squared test, Table 2).

Seventy-one of the 102 shoulders (69.6%) had a Hill-Sachs lesion and 37 (37%) had a large defect of the humeral head (>250 mm³) [[14]], which occurred at a significantly greater frequency in shoulders with re-dislocation than in those without re-dislocation (7 of 9 shoulders (78%) vs. 30 of 93 shoulders (32%), P < 0.001, chi-squared test). Significantly larger defect were also seen in the shoulders with re-dislocation compared with those without re-dislocation (834 ± 485 mm³ vs. 190 ± 255 mm³, P < 0.01, Student's t test) (Table 3).

A glenoid defect was noted in 20 of the 102 shoulders (19%) and was more prominent in the re-dislocated compared to the non-re-dislocated shoulders (4 of 9 shoulders (44%) vs. 16 of 93 shoulders (17%), P = 0.049, chi-squared test). A critical defect >20% [[18]] was noted in three dislocated and seven non-dislocated shoulders (9.8%).

Using a chi-squared test, we found that a large Hill-Sachs lesion (>250 mm³) [[5]] (P = 0.013), glenoid bone defect (>20%), and less than four suture anchors (P = 0.011) were significant risk factors for recurrence after ABR (Table 4). In contrast, there was no evidence of a relationship between re-dislocation and other factors such as age at the time of first dislocation (P = 0.27), gender (P = 0.68), the number of previous dislocations before ABR (P = 0.28), waiting time prior to surgery (P = 0.30), arm dominance (P = 0.59), injured side (P = 0.49), SLAP lesion (P = 0.27), or capsular tear (P = 0.62).

When the variables that demonstrated significance with the chi-squared test were further entered into multivariate analysis, the number of suture anchors used (odds ratio, 9.56; 95% CI, 1.99-71.4) and large Hill-Sachs lesions (odds ratio, 9.14; 95% CI, 1.90-68.3) remained independently predictive (Table 5).

No complications related to the anchors or sutures were noted in the present series, although one patient had an acute superficial infection that was readily resolved with antibiotics administration.