Modified Kouvalchouk technique for recurrent posterior instability of the shoulder

Abstract

Background

Surgical treatment of recurrent posterior shoulder instability gives good results in cases of isolated posterior labral lesion and relies on arthroscopic capsulo-labral repair. In cases of associated posterior glenoid bony lesions or for revision surgery, iliac crest bone block procedures seem to yield good short-term results, but there might be a concern for long-term results with resorption of the bone block, recurrence, and late arthritis. The Kouvalchouk technique, described in 1993, is using a bone block harvested from the posterior acromion and pedicled to a band of posterior deltoid muscle with presumed sling and vascularization effects.

Methods

The primary objective of the study was to assess the mid-term results of a slightly modified Kouvalchouk procedure (open or arthroscopic-assisted) for recurrent posterior shoulder instability. The secondary objectives were to describe the results in subgroups determined by the origin of the instability (traumatic or nontraumatic), surgical history (primary surgery or revision surgery), or surgical technique (open or arthroscopic-assisted surgery) and to evaluate the bone healing of the bone block, its position on the glenoid in the axial and sagittal planes, and the influence of bone loss and glenoid version on pre-operative and post-operative clinical data.

Results

Thirty-one procedures were followed over an average period of 4.8 years, and at the last evaluation, 77% of shoulders were stable. All bone blocks were healed. The stability rate was higher in cases of post-traumatic instability (87% vs. 66% for nontraumatic cases). Similarly, Walch-Duplay scores and patient satisfaction were better in traumatic cases (80 vs. 57 and 81% of satisfied patients vs. 40%, respectively). Previous surgery was an unfavorable condition regarding clinical results. As for clinical results or bone block positioning, arthroscopic assistance did not prove better than open surgery.Revision surgery was mostly related to recurrence and was more frequent in atraumatic cases or in previously operated patients. There were as many removals of screws for pain reasons in traumatic and atraumatic cases. Higher glenoid retroversion was associated with lower pre-operative Walch-Duplay and Rowe scores. Glenoid version lower than − 10° was associated with a higher Walch-Duplay postoperative score. Glenoid bone loss had no influence on clinical results.

Conclusion

The modified Kouvalchouk procedure provides good results in the stabilization of recurrent posterior unstable shoulders in traumatic cases and patients without previous surgery, with the advantage of local harvesting of a bone block and a potential sling effect.

Recurrent shoulder posterior instability is generally described as a rare condition, accounting for 2 to 10% of recurrent shoulder dislocations, but its frequency may be as high as 24% in young active patients.^8,30,32 Surgical treatment is an increasingly used option¹⁸ and has been shown to be preferable to medical treatment in the presence of anatomical lesions.⁵ Techniques vary and depend on types of anatomical damage to the shoulder: posterior glenoid bony and capsulo-labral lesions and anterior humeral bony lesions. They also depend on the natural condition of the shoulder (glenoid retroversion, capsular laxity), the type and level of sports practiced by patients (contact sport or not), and sometimes on the condition of the patient (epilepsy).⁸

The usual surgical techniques include soft tissue procedures, bone grafts, or a combination of both.

Among techniques using soft tissues, posterior Bankart repair with posterior and/or inferior capsulorrhaphy, either open or arthroscopic, has been widely used with good results in patients with isolated posterior labral lesions^2,8,10,25. As for techniques using bone grafts, iliac crest bone block graft has also been widely used, most of the time associated with posterior capsulo-labral repair.^{1,3,4,8,9,11,13,19,21,26,28,29} Other techniques (posterior glenoid osteotomy in case of excessive glenoid retroversion,²⁷ open or arthroscopic McLaughlin procedure³¹) are less common.

In 1993, Jean François Kouvalchouk described a technique using a fragment of posterior acromion pedicled on a band of posterior deltoid.¹⁶ The posterior bone block is screwed on the posterior glenoid, and the deltoid muscle band is supposed to act in the same way as the “sling effect” described in the Latarjet technique for the treatment of recurrent anterior dislocation. This sling effect seems to be effective compared to a single bone block, according to a recent biomechanical study.⁷

The initial technique published by Kouvalchouk¹⁶ has been modified by the main author of this study, with a vertical incision that better preserves the posterior deltoid muscle.

We more recently described an evolution of our technique²⁰ using arthroscopy, which allows an easier control of screws direction during bone block fixation and of capsulo-labral repair. This study presents the first series of patients operated on using the modified Kouvalchouk techniques (open or arthroscopic-assisted). The aim of this study was to describe their mid-term outcomes with stabilization and patient satisfaction rate as primary objectives and influence of pre-operative conditions—traumatic vs. atraumatic, primary vs. revision, surgical technique—arthroscopic-assisted vs. open, glenoid version, and glenoid bone loss—on clinical results as secondary objectives.

Material and methods

Population

This is a retrospective, cross-sectional, multicenter study aimed at evaluating the medium-term results of patients who underwent surgery for chronic posterior shoulder instability using the modified Kouvalchouk technique. The protocol was approved by the IRB (institutional review board) of the “Groupement de Coopération Sanitaire Ramsay Générale de Santé pour l’Enseignement et la Recherche” (n° COS-RGDS-2023-12-021-GRIMBERG-J).

It included 30 adult patients who were informed about the research and did not express any objection to the use of their medical data. Patients with mental disabilities or any other condition that could interfere with their understanding of the information letter, and patients under legal guardianship, could not be included.

Inclusion criteria were all patients with clinical presentation of posterior recurrent shoulder instability with conditions where soft tissue stabilization could give unsatisfactory results: bony lesions >10%, glenoid retroversion >10°, contact sports (rugby, judo), hyperlax Ehlers-Danlos disease, epileptic condition, and revision cases after failure of previous surgery (soft tissue or iliac crest bone block technique).

Exclusion criteria: primary traumatic cases with isolated soft tissue lesions and/or bony lesions <10%.

Surgical technique

A vertical 10-cm approach is drawn posteriorly along the glenohumeral joint line (Fig. 1). The posterior deltoid muscle and posterolateral part of the acromion are exposed. A 2.5 cm-long, 1.5 to 2 cm-wide, and 0.5 to 0.6 cm-thick acromial bone block is harvested, starting 2 centimeters medially from the posterolateral corner of the acromion (Fig. 2) along with a 4- or 5-cm band of posterior deltoid muscle dissected downward and laterally in alignment with muscle fibers. Only a bicortical bone block should be removed, preserving the deep part of the acromion in order to have cancellous bone on the deep portion of the bone block and to prevent subsequent fracture of the acromion (Fig. 3). Care must be taken not to separate the bone block from its deltoid attachment.

Figure 1.

Posterior view of a left shoulder pre-operatively in the operating room: pre-operative drawings for incision line (large dotted line), acromion and acromial bone block landmarks (continuous line), band of posterior deltoid muscle (small dotted line), and glenoid and humeral head landmarks (small dotted line).

Figure 2.

Acromial bone block and attached deltoid muscular band.

Figure 3.

Three-dimensional computerized tomography of a post-operative shoulder showing the screwed bone block in a correct position on the glenoid and the localization of bone harvesting with preservation of the deep bone surface of the posterior acromion.

The difference compared to the original Kouvalchouk technique is a smaller vertical approach (rather than a long horizontal one), the bicortical bone block (rather than tricortical), and the absence of detachment and reattachment of the medial portion of the posterior deltoid from the acromion. Using a vertical approach along the glenohumeral joint allows the direct approach of the joint through a horizontal dissection of the posterior cuff muscles, in the infraspinatus or between the infraspinatus and the teres minor. It is imperative to remain lateral from the distal part of the suprascapular nerve, particularly at the superior level of the joint. The posterior capsule is then divided in a horizontal T-fashion giving 2 flaps, 1 superior and 1 inferior, which may be kept with traction sutures for a subsequent posterior and inferior capsulorrhaphy. The joint is exposed, and the posteroinferior part of the glenoid bone is released from the labrum and the capsule and abraded to achieve a smooth and bleeding surface. The bone block is held on the posterior glenoid flush to the joint line and provisionally fixed with 2 guiding pins. In the open technique, one must check at this point the orientation and position of the pins for there is a risk, with limited vision inside the joint, of joint penetration. In the arthroscopic-assisted technique, this point is easily checked. Then the bone block is fixed with 2 canulated cancellous screws, diameter 3.5 or 4 (Figure 3, Figure 4). Anchors and sutures may then be used for the posterior and inferior capsulo-labral repair (Fig. 5).

Figure 4.

Arthroscopic view of the posterior aspect of the glenoid of the left shoulder after acromial bone block fixation.

Figure 5.

Same view than Figure 4 after capsulo-labral repair above the acromial bone block.

Skin and underskin are closed without any deltoid muscle repair attempt or rotator cuff tendon closure and without drainage. The shoulder is then immobilized in neutral rotation on a splint in slight abduction (20°) for three weeks and in a usual sling for 3 more weeks. Passive movements in anteflexion in the plane of the scapula start 2 days after surgery. No passive or active internal rotation or active external rotation is authorized before 6 weeks.

Objectives

The main objective of the study was to describe the medium-term results of the modified Kouvalchouk procedure in terms of posterior shoulder stability and patient satisfaction. The secondary objectives were to evaluate the influence of pre-operative conditions on post-operative results (shoulders with a clear involuntary traumatic event vs. atraumatic shoulders where trauma was minor or not found in patients' history),^8,24 surgical history (primary surgery or revision surgery), surgical technique (open or arthroscopic-assisted surgery) and to evaluate the contribution of arthroscopic assistance in placing the bone block on the glenoid and, finally, to evaluate the bone healing of the bone block, its position on the glenoid in the axial and sagittal planes and the influence of glenoid bone loss and glenoid version on the clinical results.

Assessments

The preoperative assessment included:

• •Demographic data, type of work (manual or non-manual), sport practice.
• •Disease history: initial trauma, onset of symptoms.
• •Description of symptoms: posterior dislocation needing reduction, subluxations, pain, abnormal sound during shoulder movements.
• •Active mobility in forward elevation (AF), active external rotation at the side (ER1), active external rotation at 90° of abduction (ER2), active internal rotation with the level reached by the thumb in the back (IR1).
• •Assessment of shoulder stability with the jerk test²⁵ and Kim test¹⁴: reproduction of a clear posterior dislocation, or pain with/without a clicking sensation, reproducing the patient's symptoms in cases where posterior dislocation could not be obtained.
• •Pre-operative Walch-Duplay and Rowe scores.¹⁷
• •Radiological assessment: all patients had plain X-rays of the shoulders and arthrographic computed tomography (CT) with assessment of posterior glenoid bony lesions with the best fit circle method, of glenoid version according to a modification of the Friedman method, and of the existence of posterior labral lesions and humeral lesions (reverse Hill Sachs lesion).

The follow-up assessment included

• •Description of symptoms: instability with clear episodes of dislocation or subluxation, residual pain with 3 categories: painless, rare pain with efforts, permanent pain.
• •Satisfaction toward the surgical procedure (very satisfied, satisfied, disappointed, unsatisfied).
• •Active and passive mobility.
• •Walch-Duplay and Rowe scores.
• •Radiological assessment: all patients were asked to get a postoperative CT at least 3 months after the surgical procedure and within 6 months of the index surgery to determine the position of the bone block relative to the glenoid and its cartilage surface along with bone healing (Fig. 3). Bone block and screw position was evaluated according to the criteria of Kany et al.¹² Position of the bone block in the sagittal plane was determined as correct if more than 75% of the bone block was below the equator of the glenoid and incorrect if more than 75% of the bone block was above the equator of the glenoid.
• •Complications with or without revision surgeries.

Statistical analysis

Analysis was conducted with SAS 7.04 (SAS Software, Medmenham, Marlow, Buckinghamshire, UK).

Only patients who did not object to the use of their personal data were analyzed. As this was a retrospective study with no preconceived hypotheses, the statistical analysis was purely descriptive, and the statistical tests were provided for informational purposes only, with a significance level set at P < .05. Quantitative variables were described by their frequency, mean, standard deviation, median, and extreme values and compared between 2 groups with Student's t-test or with nonparametric Mann-Whitney's tests, depending on data distribution. Comparisons of a continuous variable across more than 2 groups were conducted using the Kruskal-Wallis test. Qualitative variables were described by their frequency and percentage and compared with chi-square test or Fisher exact test. Associations between continuous variables were explored using linear regression models, and statistical significance was evaluated using the Wald test.

Results

Population

Between December 2005 and April 2021, 5 surgeons operated 40 posterior shoulders with the modified, open or arthroscopic-assisted, Kouvalchouk procedure on 39 patients (J.G. 18 cases, J.K. 10 cases, A.B. 7 cases, P.M. 4 cases, and L.B. 2 cases). Nine patients (22.5%) were lost to follow-up, leaving 31 cases available for analysis.

The study population consists of 22 men (73.3%) and 8 women. The average age at the time of the intervention was 29.7 ± 10.2 years. 15 patients (50%) were manual workers, and 19 patients (63.3%) practiced sports before the onset of symptoms.

Preoperative assessment

The assessment covered 31 shoulders: right shoulder in 21 cases (67.7%) and dominant side in 25 cases (80.6%). A posterior dislocation was documented in 87% of cases (n = 27). Two patients had recurrent subluxations, and 2 patients had painful shoulders without a clear notion of posterior dislocation or subluxation. A trauma event was found in 20 cases, and 4 of them were classified as minor. Posterior shoulder instability was, hence, classified as traumatic (n = 16) or atraumatic (n = 15).

In 15 cases (48.4%), the shoulder had already undergone one or more anterior or posterior stabilization surgeries: posterior Bankart repairs (n = 9), posterior iliac crest bone block grafts (n = 3), posterior shrinkage (n = 1), circumferential Bankart repairs (n = 2), Latarjet procedures (n = 3), or Trillat procedures (n = 1).

The clinical examination showed a clear posterior unstable shoulder with reproduction of a posterior dislocation in 19 cases (61.3%), and a posterior apprehension and/or pain and/or clicking sensation in adduction-medial rotation-retropulsion of the shoulder without dislocation in 12 cases (38.7%).

Arthro-CT showed the presence of a posterior bone and/or labral lesion in 20 cases (64.5%). Twenty-three pre-operative CT out of 31 were still available at the time of follow-up and allowed glenoid version and bone loss measurements: 7 patients out of 23 had a glenoid retroversion higher than 10°, and 15 patients had significant bone loss (10% or more). Six patients had a reverse Hill-Sachs lesion.

Average Walch-Duplay score was 19 (min −20, max 50), and average Rowe score was 31 (min 5, max 65).

Amplitude measurements performed before surgery are shown in Table I.

Table I.

Pre-operative and post-operative shoulder active mobility and post-operative clinical results (global series).

Assessments (mean ± SD)	Pre-operative (n = 31)	Post-operative (n = 31)	Paired sample t-test
Active forward flexion (°)	160.6 ± 24.8	161.3 ± 27.9	NS
Active external rotation at the side (°)	75.5 ± 19.1	69.7 ± 20.2	NS
Active external rotation in abduction (°)	91.5 ± 14.2	81.5 ± 20.0	NS
Active internal rotation (°)	9.2 ± 2.4	9.9 ± 2.8	NS
Walch-Duplay score	18.5 ± 20.4	68.9 ± 28.1	P < .001
Rowe score	30.6 ± 14.8	71.8 (±24.3)	P < .001
Patient satisfaction
Very satisfied or satisfied.		19 (61.2%)
Disappointed or unsatisfied		12 (38.8%)
Surgical revisions		11 (35.5%)

SD, standard deviation; NS, non significative.

Postoperatively

Among the 31 analyzed cases, 16 (51.6%) were openly performed, and 15 were arthroscopic-assisted (Table I).

Patients were evaluated after an average follow-up period of 74.1 ± 47.3 months (median 58 months). For the main author of this study, patients were evaluated by an independent examiner (second author) during a specific consultation or through a phone call for those who could not come back for a consultation. Patients of the other authors were evaluated by the authors themselves in the same way.

On the last follow-up, 77.4% of the shoulders operated on (n = 24) were stable.

The measurements recorded during this assessment are shown in Table I.

Overall, 19 patients (61.2%) were satisfied or very satisfied with the procedure, 6 (19.4%) were disappointed, and 6 (19.4%) were unsatisfied. Walch-Duplay score was good or very good in 16 cases (51.6%), moderate in 7 cases (22.6%), and bad in 8 cases (25.8%).

58.1% of patients experienced no postoperative pain, 12.9% reported minor pain related to efforts, and 29% felt permanent residual pain.

Post-operative CT was obtained in 25 patients and showed healing of the bone block at the glenoid in all cases. According to Kany et al,¹² post-operative bone block position on the glenoid was correct (in the sagittal plane, inferior to the equator of the glenoid) in 17 cases (Fig. 3), too high in 5 cases, too lateral (with more than 3 millimeters of lateral overhanging) in 2 cases, and medialized (with more than 3 millimeters of medialization) in 1 case.

Fifteen patients (78.9%) practicing sport before the onset of symptoms were able to resume their pre-operative activity.

Eleven patients (32%) underwent secondary surgeries: 7 patients (22.6%) had revision surgery because of a recurrence of instability, 3 patients had secondary screw removal because of residual pain without recurrence of instability, and 2 patients had fixation of acromial fracture—1 intra-operatively and 1 after a fall. One other patient had recurrence of instability but has not been revised yet. One patient had post-operative infection, which resolved after medical treatment. None of the patients had neurological complications.

Secondary criteria

Subgroup analyses were performed to compare the results in terms of stability, mobility, and functional scores of the shoulders:

• -traumatic vs. nontraumatic.
• -with a history of surgery vs. no history of surgery.
• -operated on open vs. with arthroscopic assistance.

These data are reported in Table II, Table III, Table IV.

Table II.

Results: comparison between traumatic and atraumatic cases.

Assessments (mean ± SD)	Traumatic (n = 16)	Atraumatic (n = 15)	Difference
Active forward flexion (°)	170.0 ± 13.7	152.0 ± 35.9	MW: NS
Active external rotation at the side (°)	73.1 ± 12.5	66.0 ± 26.1	MW: NS
Active external rotation in abduction (°)	83.8 ± 10.9	79.0 ± 26.7	MW: NS
Active internal rotation (°)	9.8 ± 2.3	10.1 ± 3.3	MW: NS
Walch-Duplay score	80.0 ± 21.3	57.0 ± 30.2	MW: P = .023
Rowe score	77.8 ± 21.2	65.3 ± 26.4	MW: NS
Stability rate	14 (87.5%)	10 (66.6%)	K2: NS
Pain	5 (31.2%)	5 (30.0%)	K2: NS
Patient satisfaction
Very satisfied or satisfied.	13 (81.2%)	6 (40.0%)
Disappointed or unsatisfied	3 (18.8%)	9 (60.0%)	K2: P = .04
Surgical revisions	3 (18.8%)	7 (46.7%)	K2: NS

NS, non significative; MW, Mann-Whitney test; K2, chi-square test.

Table III.

Results: comparison between primary and revision cases.

Assessments (mean ± SD)	No surgical history (n = 18)	Surgical history (n = 13)	Difference
Active forward flexion (°)	166.7 ± 22.8	153.8 ± 33.3	MW: NS
Active external rotation at the side (°)	73.1 ± 17.2	65.0 ± 23.6	MW: NS
Active external rotation in abduction (°)	83.6 ± 16.1	78.5 ± 24.8	MW: NS
Active internal rotation (°)	8.8 ± 2.3	11.5 ± 2.8	MW: NS
Walch-Duplay score	80.3 ± 20.7	53.1 ± 30.0	MW: P = .1
Rowe score	82.2 ± 19.0	57.3 ± 23.9	MW: P = .5
Stability rate	17 (94.4%)	7 (53.8%)	K2: P = .5
Pain	6 (33.3%)	4 (30.8%)	K2: NS
Patient satisfaction
Very satisfied or satisfied.	14 (77.8%)	5 (38.5%)
Disappointed or unsatisfied	4 (22.2%)	8 (61.5%)	K2: P = .027
Surgical revisions	4 (22.2%)	6 (46.2%)	K2: NS

NS, non significative; MW, Mann-Whitney test; K2, chi-square test.

Table IV.

Results: comparison between patients operated on with open or arthroscopic-assisted technique.

Assessments (mean ± SD)	Arthroscopic assisted (n = 15)	Open (n = 16)	Difference
Active forward flexion (°)	156.0 ± 31.8	166.3 ± 23.6	MW: NS
Active external rotation at the side (°)	70.0 ± 27.7	69.4 ± 10.0	MW: NS
Active external rotation in abduction (°)	76.3 ± 27.5	86.3 ± 6.2	MW: NS
Active internal rotation (°)	9.9 ± 3.4	9.9 ± 2.2	MW: NS
Walch-Duplay score	66.0 ± 26.3	71.6 ± 30.3	MW: NS
Rowe score	70.3 ± 22.9	73.1 ± 26.1	MW: NS
Stability	12 (75.0%)	12 (80.0%)	K2: NS
Pain	5 (30.0%)	5 (31.2%)	K2: NS
Patient satisfaction
Very satisfied or satisfied.	9 (60.0%)	13 (81.2%)
Disappointed or unsatisfied	6 (40.0%)	3 (18.8%)	K2: NS
Surgical revisions	5 (30.0%)	5 (31.2%)

NS, non significative; MW, Mann-Whitney test; K2, chi-square test.

There was a statistical difference in favor of traumatic cases between traumatic cases and atraumatic cases regarding patient satisfaction and post-operative Walch-Duplay score.

Although post-operative stability, pain, Rowe score, and revision surgery rates were different between the 2 groups and in favor of traumatic cases, the difference was not statistically significant.

There was a statistical difference between patients with previous surgery and patients without previous surgery, in favor of patients without previous surgery, regarding patient satisfaction, post-operative stability and post-operative Walch-Duplay and Rowe scores.

There was no difference between those 2 groups regarding surgical revisions and post-operative pain.

There was no difference regarding clinical results or regarding bone block positions between patients operated with an open technique and patients operated with an arthroscopic-assisted technique.

Higher glenoid retroversion was associated with lower pre-operative Walch-Duplay and Rowe scores. (Fig. 6). Glenoid version lower than −10° was significatively associated with a better post-operative Walch-Duplay score. Patients with glenoid version lower than −10° had a better post-operative Rowe score, but the difference was not statistically significant compared to patients with glenoid version higher than −10° (Table V). Glenoid version had no influence on post-operative stability, pain, or surgical revision. It had no relation with the mechanism (traumatic or atraumatic) of instability.

Figure 6.

Influence of glenoid retroversion on pre-operative Walch-Duplay and Rowe scores. DF, degrees of freedom; MSE, mean squared error.

Table V.

Results: binary analysis of the influence of glenoid version and glenoid bone loss on post-operative Walch-Duplay and Rowe scores.

Post-operative mean scores	Glenoid version < −10°			Glenoid bone loss >10%
	Yes	No	Difference	Yes	No	Difference
Walch-Duplay	85.0 (+/10.8)	60.0 ( ± 26.8)	Tt: P = 0,04	69.3 ( ± 26.4)	64.4 ( ± 25.7)	MW: NS
Rowe	82.1 ( ± 12.9)	65.3 ( ± 22.5)	Tt: NS (P = 0,8)	72.0 ( ± 22.3)	67.5 ( ± 20.4)	Tt: NS

MW, Mann-Whitney test; Tt, Student's t-test; NS, non significative.

Glenoid bone loss had no influence on pre-operative and post-operative scores (Table V), neither on post-operative stability, pain or surgical revision. It was not influenced by the mechanism of instability.

Finally, there was no difference in terms of stability, scores, or revision surgery between patients with a correct post-operative position of the bone block in the sagittal plane, and patients with an incorrect position. Paine was higher and satisfaction lower with patients having their bone block in an incorrect position, but with no significant difference.

Discussion

In this retrospective multicenter study of 31 cases of recurrent posterior shoulder instability, the modified Kouvalchouk procedure achieved joint stabilization in 77% of cases. However, 38.7% of patients who underwent surgery remained disappointed or unsatisfied because of residual pain or recurrence, and 35.5% of the patients underwent revision surgery for recurrence, removal of the material, or acromial fracture.

The results of the procedure varied depending on the origin of the lesions and the surgical history of the patients: post-traumatic instabilities had better results than atraumatic ones, with 87.5% of stable patients (compared to 66.7%), an average Walch-Duplay score of 80 (compared to 57), and a patient satisfaction rate of 81.2% (compared to 40%). Patients who had not undergone surgery prior to the index procedure also had better outcomes: 94.3% of patients were stable (vs. 53.8%), the mean Walch-Duplay score was 80 (vs. 53), the mean Rowe score was 82 (vs. 57) and patient satisfaction rate was 77.8% (compared to 38.5%).

Other conditions such as the learning curve of all operators (fracture of the bone block, acromial fracture, poor positioning of the bone block on the glenoid) or the specific condition of the patient (hyperlax Ehlers-Danlos syndrome confirmed in 2 patients with residual pain or recurrence) could also explain certain complications and some poor results.

We were unable to demonstrate any difference between open and arthroscopic techniques. This could be explained by the insufficient number of patients and the effect of the learning curve for both techniques.

Higher glenoid retroversion had a negative influence on pre-operative Walch-Duplay and Rowe scores, which seems logical, as one might think that a higher retroversion would be a favorable factor for posterior shoulder instability. The fact that a glenoid version less than −10° was associated with a better post-operative Walch-Duplay score might show that correction of version with the Kouvalchouk technique is effective in this specific category of patients. However, we could not prove any relation between glenoid version and glenoid bone loss and any of the following post-operative data: stability, pain, or revision surgery. The insufficient number of available pre-operative CT at the time of follow-up might be an explanation.

Surgical treatment options vary in recurrent posterior shoulder dislocation.

Open and arthroscopic Bankart along with capsular reconstruction have been widely studied: they are efficient when posterior instability is traumatic and when reparable labral and capsular lesions can be identified without bone or cartilage damage.^2,8,10,25 Atraumatic instability and/or posterior instability without well-defined capsulo-labral lesions have a less favorable prognosis.^8,23

Bone block techniques have also been widely used, but reports of results often concern series with a small number of patients and/or short-term follow-up, and in some series, questions remain regarding bone block resorption, recurrence rate, and residual pain.^{1,3,4,6,8,11,13,19,21,26,28,29}

A study from Meuffels et al even raises alarming concerns about long-term results (72% of unstable shoulders and 100% of osteoarthritis at 18 years of follow-up) with the iliac crest bone block technique overlapping the glenoid.²¹

The results of the Kouvalchouk technique have only been published in 3 articles with a limited number of patients: Kouvalchouk's study,¹⁵ involving 11 patients with an average follow-up of 5.5 years; Sirveaux's study,²⁹ where 9 patients with an average follow-up of 3.5 years were compared with 9 iliac crest bone block procedures with an average follow-up of 13.5 years; and Clavert's study,⁴ where 9 patients operated on with the Kouvalchouk procedure were retrospectively evaluated with an average follow-up of 3.3 years among 66 operated patients including 57 patients operated on with an open iliac crest bone block technique.

Our article covers a larger sample of patients. Seventy-seven percent of them were stable, compared to 82% in Kouvalchouk's study, but in the latter, only one patient had undergone a previous stabilization surgery compared to 48% in our study.

The average Walch-Duplay score after surgery in our study was 69 compared to 86 for patients in Sirveaux's study, but again, 8 out of the 9 patients operated on using the Kouvalchouk procedure in their study were post-traumatic cases (89% vs. 51% in our series), and none of them had been previously operated on.

The results in Clavert's study were not separated relatively to the 2 different techniques, so no comparison with our series was possible. Moreover, revision cases were excluded in their series. Clavert's study is the only other study in the literature comparing traumatic and atraumatic cases, but the authors did not compare the 2 categories of patients with the same clinical parameters than the ones in our study (no comparison of stability, Walch-Duplay, or Rowe scores, or patient satisfaction in Clavert's study).

Our study naturally had certain limitations: the relatively small sample size, the number of patients lost to follow-up, the absence of a control group, the absence of postoperative CT evaluation of healing and bone block positioning in some patients, the limited number of available pre-operative CT, and the absence of late postoperative CT or radiographic examination to assess bone block resorption and arthritis in the medium term.

Residual pain, when related to screws, was also a concern, but this could be resolved by modifying the bone block fixation system toward endobuttons or anchor-fixation systems, which have been found to be less problematic in terms of residual pain related to fixation material in Latarjet procedures.⁶

At the time of surgery, for most patients, assessment of the shape of the acromion according to Meyer et al²² was not available, which could explain some of the failures for atraumatic cases.

However, to date, among studies using a bone block technique, this is one of only 3 series^4,19 involving the largest number of patients, one of the 2 series⁴ with distinct results regarding mechanism of instability (traumatic vs. atraumatic), and the only one comparing results between primary and revision cases.

Moreover, this technique has a clear advantage over the iliac crest bone block technique, which is the local site for harvesting of the bone block. It might also have offered potential advantages with the sling effect⁷ and vascularization effect of the deltoid muscle, but this latter hypothesis remains to be proven by a long-term comparative evaluation of bone block resorption with other bone block techniques.

The evolution toward an arthroscopic-assisted technique might also decrease misplacement of bone block, probably responsible for some of our failures, and ease the associated capsulo-labral reconstruction, although the learning curve and limited number of patients for all the authors possibly prevented us to prove those points. Nevertheless, arthroscopic control remains a good way to prevent joint penetration when screws are used for bone-block fixation.

Comparative studies with a higher number of patients and longer follow-up will be needed in the future, even if the organization of comparative controlled studies taking into account all the influencing parameters on the results would be difficult in this rare pathological condition.

Conclusion

In a multicenter retrospective study of 31 patients, the modified Kouvalchouk procedure provided good results in the stabilization of recurrent posterior unstable shoulders. In traumatic cases and in patients without previous surgery, with the advantage of local harvesting of a bone block and potential sling effect. Patient satisfaction rate remains lower in atraumatic and revision cases where risks of residual pain or instability recurrence must be discussed preoperativly with the patients arthroscopic-assisted procedure could not be proven superior to open procedure, whether it be for clinical results or bone block positioning.