Open Bankart Repair for the Treatment of Anterior Shoulder Instability without Substantial Osseous Glenoid Defects

Abstract

Background:

Neglected osseous glenoid defects are thought to be one of the reasons for the reported high rates of recurrent instability at long-term follow-up after Bankart repair. We hypothesized that open Bankart repair for the treatment of anterior glenohumeral instability in the absence of a substantial osseous glenoid defect would result in a lower long-term recurrence rate than has been reported in previous long-term studies.

Methods:

Forty-seven patients were treated with a primary modified open Bankart repair for recurrent anterior shoulder instability between 1989 and 1994. Double-contrast computed tomography scanning was used to exclude patients with a substantial osseous glenoid defect. Forty patients (85.1%) were available for subjective and objective follow-up at a minimum of twenty years (maximum, twenty-five years). Twenty-six patients (65%) underwent clinical examination as well as bilateral shoulder radiography, and fourteen (35%) completed a self-assessment questionnaire and were interviewed by telephone.

Results:

Seven patients (17.5%) had a recurrence of instability, and six of them had the instability occur after more than eight years without symptoms. The mean Western Ontario Shoulder Instability Index score (and standard deviation) was 256.7 ± 284.8 points; the mean Rowe score, 88.7 ± 12.0 points; and the mean Subjective Shoulder Value, 90.1% ± 10.5%.The mean range of motion of the affected shoulder was decreased by 4° of abduction (p = 0.009), two levels of internal rotation (p = 0.003), 5° of internal rotation in 90° of abduction (p = 0.005), 7° of external rotation in neutral position (p < 0.001), and 7° of external rotation in 90° of abduction (p = 0.004) compared with the contralateral side. The collective instability arthropathy (CIA) index was 0.92 for the affected side and 0.35 for the contralateral side.

Conclusions:

Open Bankart repair provides good results twenty years after surgery in terms of subjective and objective outcome measurements. However, the long-term failure rate remains high despite the exclusion of substantial osseous glenoid defects. Recurrence of instability seems to be associated with an increased shoulder-specific activity level.

Level of Evidence:

Therapeutic Level IV. See Instructions for Authors for a complete description of levels of evidence.

In 1923, Bankart et al. described an open surgical technique to treat recurrent anterior shoulder instability by means of an anterior capsulolabral repair¹. Over the last decades, a number of modifications, open and arthroscopic, have been reported, but the main idea of the capsulolabral repair to stabilize the shoulder joint remained the same. While the short-term recurrence rate after a Bankart repair is generally low^2-6, the reported recurrence rates at a minimum of ten years are surprisingly high, ranging from 9.1% to 57.6%^7-11. One explanation might be the fact that prior long-term studies did not utilize computed tomography (CT) scans to exclude patients with larger osseous defects, as defined according to today’s standards^12-14, which would be treated with glenoid reconstruction by means of a bone graft rather than a Bankart procedure.

As recurrent anterior shoulder instability often affects younger individuals, it is very important to consider the long-term outcome. In addition to recurrent instability, instability arthropathy is a major concern. There are few twenty-year follow-up studies regarding postoperative instability arthropathy after Bankart repair^15,16.

The purpose of our study was to determine the minimum twenty-year clinical and radiographic follow-up results in a cohort of patients who underwent open Bankart repair after previous exclusion of those with substantial osseous glenoid defects by means of double-contrast CT scans.

Materials and Methods

The present investigation was a retrospective study with current follow-up and was approved by the local ethics committee.

Between the end of 1989 and the beginning of 1994, forty-seven consecutive patients with recurrent anterior shoulder instability were treated at the institution of the senior author (Peter Moroder) by means of an open modified Bankart repair. Of the forty patients available for follow-up, ten were female and thirty were male and they had a mean age (and standard deviation) of 26.7 ± 9.4 years at the time of surgery. The right shoulder was affected in twenty-seven patients and the left shoulder, in thirteen patients; the dominant arm was affected in 65.0% of the patients. The mean time between the first instability episode and the surgical intervention was 4.4 ± 6.5 years, with a mean of 9.2 ± 8.1 dislocations (range, two to forty dislocations) preoperatively. None of the patients had undergone a previous attempt at surgical stabilization. The first dislocation was traumatic in all patients, including 77.5% that resulted from sports incidents; 10.0%, from motor vehicle accidents; 7.5%, from a simple fall; and 5.0%, from work-related accidents. Eight patients (20%) had at least one instability event involving the contralateral shoulder, and six underwent stabilization surgery for the contralateral shoulder. Of the patients who were examined radiographically at the time of follow-up, 23% had contralateral shoulder instability and 14% underwent an open or arthroscopic Bankart procedure on the contralateral shoulder.

Prior to surgery, all patients underwent a double-contrast CT examination (Fig. 1) according to the reported protocol¹⁷ in order to exclude those with critical-sized osseous glenoid defects and Hill-Sachs lesions. Resch et al. defined a critical-sized glenoid defect as one that reduced the ratio between the largest anteroposterior glenoid concavity diameter and the largest humeral head diameter in the transverse plane below 58%¹⁸. A critical-sized Hill-Sachs lesion was defined as a defect whose margin reached farther medially than 190° measured from the anatomic neck and the midpoint of a best-fit circle in the transverse plane. In the case of a critical-sized glenoid defect, an iliac-crest bone graft would have been performed instead of the open Bankart repair, and in the presence of a large Hill-Sachs lesion, elevation of the impression fracture would have been indicated¹⁸. None of the forty-seven patients had critical-sized glenoid bone loss or a large Hill-Sachs lesion on preoperative double-contrast CT scans.

Fig. 1.

An example of a preoperative sequence of double-contrast CT scans of one of the patients.

Despite the typically short, ten-year archiving period for radiographic images in our institution, ten CT scans could be retrieved from the institutional archives. The absence of critical-sized osseous glenoid defects was reconfirmed in all ten patients, including five of the seven with postoperative recurrence of instability. However, no modern measurement technique (e.g., three-dimensional en face views and defect area measurements) of glenoid bone loss could be used retrospectively as only axial, unilateral hard-copy images were available.

The patients were treated with open Bankart repair, as modified according to the method of Resch et al., which consists of a transosseous mattress-stitch reinsertion of the capsule at the anterior glenoid rim¹⁸ (Fig. 2).

Fig. 2.

Schematic drawings of the modified open Bankart repair technique. Fig. 2-A After a vertical incision of the anterior capsule (C), a small notch (N) is created at the anterior glenoid rim (G) at the cartilage-bone junction using a chisel, and four drill-holes are made from lateral to medial starting from the notch. Figs. 2-B, 2-C, and 2-D Sutures are passed through each drill-hole, and the most inferior suture is passed through the lateral part of the incised capsule in the form of a mattress stitch and is returned through the second inferior hole. Fig. 2-E The knot is tied medially. Another mattress stitch is completed using the two superior holes. Afterward, the medial part of the incised capsule is sutured over the lateral capsule, which was previously attached to the glenoid rim.

Clinical Follow-up Evaluation

Multiple attempts were made to contact all forty-seven patients either by telephone or by mail to invite them to return for clinical and radiographic follow-up examination. Forty patients (85.1%) could be reached and agreed to be evaluated. Twenty-six patients (55.3%) were examined clinically with the Western Ontario Shoulder Instability Index (WOSI), Rowe score, Subjective Shoulder Value (SSV), visual analog scale (VAS) for pain, and range of motion determined by a goniometer, as well as abduction strength determined by a dynamometer in 90° of abduction and rotation strength in 0° of abduction^19-21. Additionally, the Beighton score for generalized joint hyperlaxity and an anterior apprehension test were assessed at the time of follow-up²². Patients had to ascribe their satisfaction level with the operatively treated shoulder to one of the following four categories: very satisfied, satisfied, indifferent, and not satisfied. Additionally, all patients had rated the degree of shoulder-related impairment during work or sports activities as none, mild, moderate, or complete. Bilateral radiographs were made in the anteroposterior and axial directions. Fourteen (29.8%) of the forty-seven patients completed a combined self-assessment questionnaire including photographic illustrations paired with a telephone interview, simulating as well as possible the above-mentioned clinical evaluation. No radiographic follow-up could be obtained for the fourteen patients. Five patients (10.6%) could not be reached because of missing updates to their contact information, and two patients (4.3%) had died of an unrelated cause. The average follow-up period was twenty-two years (range, twenty to twenty-five years).

Shoulder Activity Level (SAL)

To categorize the overall shoulder-specific exertion level of the participants, the work and sports-related shoulder activity was rated from 0 to 2 points each and was summed into a cumulative rating scale from 0 to 4 points, defined as the SAL. Work without the need for heavy labor accounted for 0 points, work requiring heavy labor but no overhead activity was rated as 1 point, and work requiring heavy labor including overhead activity accounted for 2 points. No regular athletic activity was rated as 0 points, regular athletic activity without overhead activity accounted for 1 point, and regular overhead athletic activity accounted for 2 points. Accordingly, an SAL of 0 was equated with no shoulder-specific activity; 1, with minor shoulder-specific activity; 2, with moderate shoulder-specific activity; 3, with major shoulder-specific activity; and 4, with extreme shoulder-specific activity.

Collective Instability Arthropathy (CIA) Index

Instability arthropathy was classified according to Samilson and Prieto²³ on true anteroposterior radiographs as grade-0 to grade-3 osteoarthritis. In order to allow for easier comparability with the contralateral side and with the results of other long-term studies, the CIA index was employed. For each instability arthropathy grade, a corresponding number of points between 0 and 3 was assigned. The CIA index represents the sum of all calculated points divided by the total number of examined patients. The index ranges from 0, meaning none of the patients showed a sign of instability arthropathy, to 3, meaning all patients featured a grade-3 instability arthropathy. Between 0 and 3, the CIA index does not offer a distinction between quantity and quality, meaning that, for example, of 100 patients, thirty with grade-1 instability arthropathy render the same index (0.3) as ten patients with grade-3 arthropathy.

Statistics

Descriptive statistics, including the mean, standard deviation, and minimum and maximum values of the variables, were calculated. All variables were tested for normal distribution using the Kolmogorov-Smirnov test. Normally distributed variables were compared by means of the Student t test and non-normally distributed data, using the Mann-Whitney U test. For comparison of categorical data, the chi-square test was used. To analyze correlations between variables, the Pearson correlation coefficient was calculated. The entire statistical analysis was completed with two-tailed p values and with the alpha level set at 0.05.

Source of Funding

No funding was obtained for this study.

Results

Functional Outcome

The mean reported WOSI score was 256.7 ± 284.8 points (range, 3 to 1117 points). The mean Rowe score was 88.7 ± 12.0 points (range, 49 to 100 points), and the mean SSV was 90.1% ± 10.5% (range, 55% to 100%). The mean range of motion of the index shoulder was significantly decreased by 4° in terms of abduction (p = 0.009), by two vertebral levels for internal rotation (p = 0.003), by 5° of internal rotation in 90° of abduction (p = 0.005), by 7° of external rotation in neutral position (p < 0.001), and by 7° of external rotation in 90° of abduction (p = 0.004) compared with that on the contralateral side (Table I). The mean strength of the index shoulder compared with the contralateral side reached 92.8% ± 16.6% in abduction (p = 0.014), 94.9% ± 12.3% in internal rotation (p = 0.014), and 88.6% ± 20% in external rotation (p = 0.002). The mean reported level of pain was 0.8 ± 1.2 point (range, 0 to 4 points) on the 10-point VAS. Of the forty patients, 82.5% were very satisfied with the outcome of the surgery, 15.0% were satisfied, and 2.5% felt indifferent. Seventy percent experienced no shoulder-related impairment during work; 27.5%, a mild impairment; and 2.5%, a moderate impairment. Regarding sports activity, 67.5% reported no impairment; 27.5%, a mild impairment; and 5.0%, a moderate impairment. The mean SAL of the study group at the time of follow-up was 1.7 points (range, 0 to 4 points), with seven patients reporting an SAL of 0 point; nine, an SAL of 1 point; thirteen, an SAL of 2 points; ten, an SAL of 3 points; and one, an SAL of 4 points.

TABLE I.

Comparison of the Mean Range of Motion of the Involved Shoulder and the Contralateral Side at the Time of Follow-up

	Involved Shoulder	Contralateral Shoulder	P Value
Abduction (deg)	169	173	0.009
Flexion (deg)	172	173	0.086
Internal rotation (vertebral level reached by thumb)	T9	T7	0.003
External rotation (deg)	57	64	<0.001
Internal rotation (90° of abduction) (deg)	74	79	0.005
External rotation (90° of abduction) (deg)	73	80	0.004

Recurrence of Instability

Recurrent instability during the follow-up period occurred in seven (17.5%) of the forty patients, with five patients sustaining a complete dislocation requiring reduction by a physician and two patients describing an instability episode consistent with subluxation. Six of the seven patients experienced their first postoperative instability episode after more than eight years without symptoms. Three late failures were the result of a high-energy sports injury. Two late failures occurred because of a minor trauma, and one occurred even without reported trauma (Table II). Three of the twenty-six patients who had a clinical examination featured a positive apprehension test without reporting any instability episodes.

TABLE II.

Patients Sustaining an Episode of Instability After Open Modified Bankart Repair

Case	Sex	Age at Time of Failure (yr)	Time Between Surgery and Redislocation (yr)	Cause of Redislocation	Revision Surgery
1	M	37	16	Fall from a standing position	Arthroscopic Bankart
2	M	43	12	Paragliding accident	Arthroscopic Bankart
3	F	34	17	Skiing accident	None
4	M	39	17	Inadvertent arm movement	None
5	M	20	0.5	Skiing accident	None
6	F	41	17	Fall from a standing position	None
7	M	31	8	Skiing accident	None

The SAL score of patients reporting a recurrence of instability was significantly greater than that of the other patients (p < 0.05) (Table III). The mean Beighton score at the time of follow-up was 1.0 point (range, 0 to 6 points) at the time of follow-up. There was no association between the Beighton score and recurrence of instability.

TABLE III.

Differences Between Patients with and without Redislocation After Surgery

Variables	Patients with Redislocation (N = 7)	Patients without Redislocation (N = 33)	P Value
Age at time of surgery* (yr)	22.1	27.6	0.181
Age at follow-up* (yr)	44.4	49.6	0.268
No. of dislocations	10.7	8.9	0.519
Time before surgery* (yr)	4.0	4.5	0.090
Beighton score* (points)	1.6	0.8	0.273
Dominant arm involved (no. of patients)	5	21	0.859
WOSI score* (points)	328.3	241.5	0.465
Rowe score* (points)	81.4	90.3	0.190
SSV* (%)	83.7	91.4	0.094
Abduction* (deg)	169	169	0.612
Flexion* (deg)	172	172	0.893
External rotation* (deg)	62	56	0.535
Internal rotation* (vertebral level reached by thumb)	T7	T9	0.161
External rotation in 90° of abduction* (deg)	76	73	0.599
Internal rotation in 90° of abduction* (deg)	70	74	0.372
VAS* (points)	0.6	0.8	0.757
Satisfaction* (max., 5 points) (points)	4.7	4.8	0.434
Work impairment* (max., 4 points) (points)	0.7	0.2	0.061
Sports impairment* (max., 3 points) (points)	0.6	0.3	0.185
SAL*	2.7	1.5	0.011†
CIA index*	1.2	0.9	0.677

^*Values are given as the mean.

^†Significant difference.

Instability Arthropathy

The grade of instability arthropathy observed in the affected shoulder and the contralateral side in the twenty-six patients who had a radiographic examination was 0 for thirteen and nineteen shoulders, respectively; 1 for six and six shoulders; 2 for three and zero shoulders; and 3 for four shoulders and one shoulder. The CIA index was 0.92 for the affected side and 0.35 for the contralateral side. An increase in the degree of instability arthropathy correlated significantly with a reduction in the Rowe score (p = 0.039, r = −0.424) and SSV (p = 0.032, r = −0.439), with a diminution of the external rotation at 0° (p = 0.044, r = −0.415) and 90° of abduction (p = 0.031, r = −0.451), and with an increase in work (p = 0.015, r = 0.492) and sports impairment (p = 0.008, r = 0.530) (Table IV).

TABLE IV.

Instability Arthropathy Correlations

	Correlation Coefficient (r value)	P Value
Age at time of surgery	0.211	0.322
Age at time of follow-up	0.187	0.381
No. of dislocations	−0.008	0.970
Years before surgery	−0.237	0.264
Beighton score	0.098	0.648
Osteoarthritis in contralateral shoulder	0.218	0.418
Bilateral instability	−0.184	0.391
Dominant side affected	−0.158	0.461
Failed stabilization	0.092	0.670
WOSI	0.117	0.585
Rowe	−0.424	0.039*
SSV	−0.439	0.032*
Abduction	−0.090	0.676
Flexion	0.082	0.703
External rotation	−0.415	0.044*
Internal rotation	0.399	0.053
External rotation (90° of abduction)	−0.451	0.031*
Internal rotation (90° of abduction)	−0.270	0.213
VAS	0.065	0.764
Satisfaction	−0.100	0.642
Work impairment	0.492	0.015*
Sports impairment	0.530	0.008*
SAL	−0.267	0.207

^*Significant difference.

Discussion

The goal of the present study was to evaluate the long-term clinical outcome and rate of recurrent instability after a modified open Bankart repair performed to treat patients with recurrent anterior glenohumeral dislocation without substantial osseous glenoid defects. Additionally, the occurrence of instability arthropathy in the affected shoulder and the contralateral side was analyzed.

Despite the good objective and subjective clinical outcome with a minimal loss in range of motion twenty years or more after surgery, the rate of recurrent instability of 17.5% appears similarly high compared with reports of other long-term studies, even though a double-contrast CT scan was used to exclude critical glenoid bone loss^{7-11,15,16,24-28} (Table V). The fact that six of the seven patients with failed stabilization experienced their first redislocation after more than eight years without symptoms suggests that the risk for recurrence after a Bankart repair persists over time. According to our data, there was an association between the shoulder-specific activity level of a patient and the risk for recurrence of instability, and three late failures occurred during a high-energy sports accident. Castagna et al. reported similar findings regarding late failures in patients involved in contact sports or overhead activities after arthroscopic Bankart repair²⁵. However, it remains unclear why two late failures occurred because of a minor trauma and one occurred even without reported trauma. It is possible that a lifestyle with high demands on the shoulders leads to weakening of the Bankart repair over time because of repetitive stress of the anterior capsulolabral complex. Nevertheless, recurrence of instability did not appear to significantly affect the subjective and objective outcome scores or the degree of work and sports impairment compared with the rest of the patients.

TABLE V.

Overview of the Literature Regarding Long-Term Recurrence of Instability and the Collective Instability Arthropathy (CIA) Index After Bankart Repair

Study	Technique	Minimum Follow-up (yr)	No. of Shoulders	CIA Index*	Recurrence Rate (%)
≥20 years of follow-up
Pelet et al.15 (2006)	Open Bankart	20.3	30	1.27	10.0
Fabre et al.16 (2010)	Open Bankart	25	50	1.05	16.0
Current study	Open Bankart	20	40	0.92	17.5
<20 years of follow-up
Gill et al.26 (1997)	Open Bankart	8	60	NA	5.0
Cheung et al.7 (2008)	Open Bankart	10.7	34	NA	11.8
Salomonsson et al.8 (2009)	Open Bankart	10	32	NA	57.6
Berendes et al.9 (2007)	Open Bankart	10	31	0.39	22.6
Hovelius et al.10 (2011)	Open Bankart	13.5	88	NA	27.3
Zaffagnini et al.11 (2012)	Open Bankart	10	33	0.70	9.1
Zaffagnini et al.11 (2012)	Arthroscopic Bankart	10	49	0.53	12.2
Owens et al.24 (2009)	Arthroscopic Bankart	9.1	40	NA	37.5
Castagna et al.25 (2010)	Arthroscopic Bankart	9.8	31	0.48	22.6
Kavaja et al.27 (2012)	Arthroscopic Bankart	11	81	0.82	22.9
Privitera et al.28 (2012)	Arthroscopic Bankart	10.8	20	1.35	30.0

^*NA = not available.

The CIA index of the affected shoulders in the patients who had radiographic examination was almost three times as high as that of the contralateral shoulders. An increase in the degree of instability arthropathy correlated with worse subjective and objective outcome scores as well as with increased work and sports impairment. Interestingly, instability arthropathy was associated with a loss of external rotation, raising the question of whether the loss of external rotation was caused by osteoarthritis or by overtightening of the anterior capsule. In a previous in vitro investigation, Ahmad et al. suggested that anterior tightening causes early-onset osteoarthritis²⁹. A clear cause-and-effect relationship cannot be determined on the basis of our data. With the data available, we did not find a correlation between the grade of instability arthropathy and the SAL. The CIA index of 0.92 is a little lower than the indexes of 1.1 and 1.3 reported in other twenty-year follow-up studies after open Bankart repair^15,16 but higher than the indexes of 0.4 and 0.7 reported for ten-year follow-up studies of open Bankart repair^9,11. The CIA indexes reported ten years after arthroscopic Bankart repair have had a greater range, from 0.5 to 1.4^{11,24,25,27,28} (Table V).

Our study had several limitations other than the relatively small number of patients. We did not have access to all of the preoperative CT scans and therefore could not reevaluate the glenoid bone loss. As a result, we relied on the initial evaluation made by the treating surgeon and the radiologist over twenty years earlier, according to the above-mentioned measurement method¹⁸. Nevertheless, the CT scans of five of the seven patients with failed stabilization were available, and none demonstrated a substantial osseous defect. This finding makes it highly unlikely that critical glenoid bone loss was the cause of the high rate of recurrence. Although differences in the method of determining glenoid bone loss may limit comparison with more current investigations, our method still enabled us to identify patients with larger defects. Another weakness of the present study is the fact that, despite a serious effort to achieve a follow-up rate of 85% (forty of forty-seven patients) after a minimum follow-up of twenty years, bilateral radiographs could be obtained for only 55% (twenty-six) of the patients, which limits the instability arthropathy analysis.

Another limitation is the fact that the SAL and the CIA index are nonvalidated measurement scales. Despite the above-mentioned lack of discrimination between the number of arthropathy cases and the severity of osteoarthritis encountered, we found the CIA index to be a simple and interesting tool for an approximate comparison of instability arthropathy rates reported in different long-term studies.

In conclusion, open Bankart repair provides good results twenty years after surgery in terms of subjective and objective outcome measurements. However, the long-term failure rate remains high despite the exclusion of extensive osseous glenoid defects. Recurrence of instability seems to be associated with an increased shoulder-specific activity level.