fu

dem

4

Posterior Shoulder Instability: Does Glenoid RetroversionPredict Recurrence and Contralateral Instability?

Michael B. Gottschalk, M.D., Alex Ghasem, B.A., Dane Todd, M.D.,Jimmy Daruwalla, M.D., John Xerogeanes, M.D., and Spero Karas, M.D.

Purpose: To determine whether glenoid retroversion is a predictor of posterior shoulder instability, contralateral insta-bility, or recurrent instability in patients with traumatic, contact-related posterior shoulder instability. Methods: Patientswho underwent shoulder stabilization by 2 senior orthopaedic sport surgeons were identified retrospectively. Patients witha connective tissue disorder, multidirectional instability, or nonetrauma-induced pathology were excluded. Patients witha glenoid lesion involving greater than 25% of the glenoid or an engaging humeral lesion were also excluded. Thuspatients with a traumatic injury and a magnetic resonance imaging scan available for review were included. Magneticresonance imaging scans were reviewed, and glenoid version was measured using the glenoid vault method. Charts werereviewed for epidemiologic data, recurrent instability requiring reoperation, evidence of glenoid/humeral bone lesions,and contralateral shoulder instability requiring surgery. Both recurrence and contralateral injury were defined based onhaving repeat surgery. Results: We identified 143 patients who met the inclusion criteria. Twenty-eight patients hadposterior instability, whereas 115 patients had anterior instability. Patients with posterior instability had significantlymore glenoid retroversion than patients with anterior instability (�15.4� � 5.14� v �12.1� � 6.9�; P < .016). Patientswith retroversion of more than �16� showed a higher incidence of contralateral injuries (P < .036). However, nodifference in postsurgical recurrent instability was noted. Conclusions: Our data show that patients with posteriorinstability have a higher incidence of having a retroverted glenoid. Patients with increased retroversion showedincreased posterior contralateral instability. Furthermore, patients with posterior instability and no humeral bone lesionsmay be more likely to incur contralateral injuries than those with humeral lesions. These data suggest that glenoidversion and concomitant injury patterns may be used to help physicians counsel patients on their future risks ofcontralateral injury. Level of Evidence: Level IV, therapeutic case series.

houlder instability remains one of the most

Scommonly treated entities in orthopaedic surgery.Although several studies have investigated the inci-dence and risk factors involved in anterior shoulderinstability, only recently have studies investigated theincidence of, and risk factors involved in, anteriorshoulder instability in young athletes.1,2 Because ante-rior instability and multidirectional instability (MDI) aremore common, most of the available literature pertainsto these entities. Nevertheless, studies have begun to

From Emory Orthopaedics, Atlanta, Georgia, U.S.A.The authors report the following potential conflict of interest or source ofnding: J.X. and S.K. receive support from DJO, Arthrex, ConMed.Received March 18, 2014; accepted October 2, 2014.Address correspondence to Michael B. Gottschalk, M.D., Emory Orthopae-

ics, 59 Executive Park Dr S, Atlanta, GA 30329, U.S.A. E-mail: mbgotts@ory.edu� 2015 by the Arthroscopy Association of North America0749-8063/14224/$36.00http://dx.doi.org/10.1016/j.arthro.2014.10.009

88 Arthroscopy: The Journal of Arthroscopic and Related S

investigate the risk factors, outcomes, and surgicaltreatments associated with posterior instability.Glenohumeral anatomy is one of many risk factors

studied in shoulder instability. Because of the largevariations in glenoid anatomy, studies have investi-gated the morphology and size of the labrum and gle-noid version.3-5 Whereas glenoid retroversion has beenimplicated in patients with posterior MDI, few studieshave been able to corroborate this in patients withtraumatic/athletic posterior instability.6-9

Multiple characteristics of the glenoid have beenstudied; so too has the optimal way to measure theseaspects of glenoid anatomy. Two- and 3-dimensionalmagnetic resonance imaging (MRI)/computed tomog-raphy (CT) scans comprise the majority of radiographictools used.10-22

Although glenoid version and its relationship toshoulder instability have been previously examined,few if any studies have investigated the influence ofglenoid version on contralateral shoulder instability,postoperative recurrent instability, or the direction of

urgery, Vol 31, No 3 (March), 2015: pp 488-493

Fig 1. (A) Centering of glenoid using sagittal, coronal, and axial planeelinked cross sections. A left shoulder is shown with thepatient lying supine. (B) Coned view of an axial MRI slice of a left shoulder depicting measurement of glenoid version. Lines 1and 2 show the isosceles triangle with a single line bisecting the triangle. Line 4 is a line perpendicular to the bisecting line, andline 5 is drawn in relation to the endosteal face. The glenoid version is the angle between line 4 and line 5.

POSTERIOR SHOULDER INSTABILITY 489

instability in trauma/contact athletes. The purpose ofthis study was to determine whether glenoid retrover-sion is a predictor of posterior shoulder instability,contralateral instability, or recurrent instability in pa-tients with traumatic, contact-related posterior shoul-der instability. We hypothesized that increased glenoidretroversion would be a predictor of posterior shoulderinstability, increased risk of contralateral shoulderinstability, and ipsilateral recurrence after treatment.

MethodsAfter institutional review board approval was obtained

for this study, patients who underwent arthroscopicshoulder stabilization between 2006 and 2012 by 2 or-thopaedic sports surgeons (J.X., S.K.) were identified.The inclusion criteria comprised the following: traumaticcause, age older than 14 years, MRI available for review,glenoid involvement of less than 25%, and non-engaging Hill-Sachs deformity. Patients with a connec-tive tissue disorder, MDI, or nonetrauma-inducedpathology were excluded. Patients with nontraumaticpathology were defined as those with instability eventswithout a known cause or impetus (e.g. rolling over in

bed). Patients with a glenoid lesion involving greaterthan 25% of the glenoid or an engaging humeral lesionwere also excluded. One hundred forty-three patientsmet the inclusion criteria and had MRI scans availablefor review. Before any surgical intervention, each pa-tient underwent an examination under anesthesia bythe attending surgeon, which included the load-and-shift test. Patients were categorized as having eitheranterior or posterior instability based on the attendingsurgeon’s postoperative diagnosis as listed in the oper-ative report.A single reviewer (A.G.) examined all preoperative

MRI scans to measure glenoid version. Exact versiondifferences were �15� and �12�. All MRI scans wereperformed on the same scanner with axial images.Formatted images, including coronal and sagittalreformatted images, were also available. Glenoidversion was measured using the glenoid vault methodas discussed by Poon and Ting.10 Figure 1 shows anexample measurement. First, an isosceles triangle isdrawn within the glenoid cavity. Then, a line bisectingthe triangle is drawn, and a line perpendicular to thisbisector is drawn. Finally, a line paralleling the glenoid

Table 1. Epidemiologic Data by Group

Posterior Anterior

P Valuen Mean SD Lower 95% Upper 5% n Mean SD Lower 95% Upper 5%

Age, yr 28 23.89 9.64 20.16 27.63 115 26.55 9.24 24.84 28.25 .038Follow-up, mo 28 11.54 6.05 17.02 115 8.06 5.12 10.99 .013Male 27 87 .01Female 1 28 .01Glenoid version 28 �15.4� 5.14� �17.39� �13.41� 115 �12.12� 6.9� �13.4� �10.85� .016

490 M. B. GOTTSCHALK ET AL.

endosteal face is drawn, which is then compared withthe perpendicular line of the bisector to give the glenoidversion. Version was measured a second time approx-imately 3 weeks later to help address the variability inmeasuring glenoid version. These measurements wereaveraged to obtain a single version value for each pa-tient. Of note, Poon and Ting used CT scans for evalu-ation of glenoid version in their study whereas ourstudy used MRI. MRI was used because of the avail-ability of MRI scans versus CT scans for the patientsreviewed in our study (CT scans are routinely reservedfor those patients with large bony lesions or glenoidhypoplasia). This method for glenoid version evaluationwas selected because of its high interobserver andintraobserver reliability as shown by Poon and Ting.Epidemiologic data, the presence of glenoid or humeralbone lesions (diagnosed at the time of arthroscopy),reinjury or failure after surgery (defined as the need forrepeat surgery), and contralateral instability (defined asthe need for contralateral surgery because of instability)were all recorded. Of note, because our institution is atertiary referral center, most of the patients presented tothe clinic after sustaining multiple instability events. Inthe case of a single event, the decision to perform sur-gery or repeat operation is individualized based on adiscussion with the patient on the basis of his or herage, activity level, and want or need to return to sport.The nonparametric Wilcoxon rank sum test and

Mann-Whitney U test were used for continuous

Table 2. Subset Analyses of Variables

Gender Humeral Bone Lesion

Male FemaleP

Value Yes NoP

Value

Contralateral injury .57 .48Yes 19 3 10 11No 94 26 67 50

Glenoid lesion .48 .035*Yes 32 6 27 11No 78 23 51 50

Humeral lesion .84Yes 61 17No 49 12

Reinjury/reoperation >.99 >.99Yes 13 3 8 6No 100 26 69 55

Glenoid version �12.5� �13.9� .52 �12.8� �12.6� .73

*Statistically significant.

variables, whereas the c2 and Fisher exact tests wereused for categorical/nominal data when the number ofpatients was greater than 10 and less than 10, respec-tively. A P value of .05 was set for statisticalsignificance.

ResultsEpidemiologic data were available for all 143 patients.

The mean age was 23.89 � 9.64 years (95% confidenceinterval [CI], 20.16 to 27.63 years) and 26.55 � 9.24years (95% CI, 24.84 to 28.25 years) in the posteriorand anterior groups, respectively. The posterior insta-bility group showed a statistically significantly youngerpatient population (P ¼ .038). There were 27 malepatients and 1 female patient in the posterior group and87 male and 28 female patients in the anterior group.There was a significantly higher percentage of malepatients in the posterior group: 96% versus 76% (P ¼.01). The mean follow-up time was 11.54 months (95%CI, 6.05 to 17.02 months) and 8.06 months (95% CI,5.12 to 10.99 months) in the posterior and anteriorgroups, respectively. Follow-up was significantly longerin the posterior group (P ¼ .013). Although both ageand gender differed significantly in each group, onlyage showed a statistically significant differenceregarding glenoid version when multivariate and uni-variate analyses were performed (P ¼ .005). Handdominance and the side requiring surgery were evalu-ated, and no trends or differences were noted between

Glenoid Bone Lesion Age

Yes NoP

Value <20 yr 20-24 yr 25-30 yr >30 yr P

.29 .828 13 5 6 4 7

30 87 34 39 18 29.28

10 9 6 1329 37 15 20

.035* .5827 51 19 25 13 2111 50 20 21 8 12

>.99 .314 10 6 2 3 5

34 90 33 43 19 31�14.1� �12.2� .21 �12.4� �9.85� �16� �14.7� .005*

Table 3. Injury Pattern by Subgroup Variables

Posterior Anterior

P ValueYes No Yes No

Contralateral surgery (n ¼ 142)(1 lost to follow-up)

9 19 13 101 .01

Glenoid bone lesion (n ¼ 139) 6 22 32 79 .49Humeral bone lesion (n ¼ 139) 5 23 73 38 .001Reinjury requiring reoperation

(n ¼ 142) (1 lost to follow-up)2 26 14 100 .74

POSTERIOR SHOULDER INSTABILITY 491

groups. Epidemiologic data are summarized in Table 1.Subset analysis is shown in Table 2, and injury patternsby subgroup are shown in Table 3. Mechanisms ofinjury for the primary shoulders, contralateral shoul-ders, and shoulders with reinjury are summarized inTables 4, 5, and 6, respectively.Glenoid version was �15.4� � 5.14� (95%

CI, �17.39� to �13.41�) and �12.12� � 6.9� (95%CI, �13.4� to �10.85�) in the posterior and anteriorgroups, respectively. The posterior group had signifi-cantly more retroverted glenoids than the anteriorgroup (P ¼ .016). These data are also summarized inTable 1.The incidence of contralateral instability requiring

surgery was available for 142 of the 143 patientsreviewed (1 patient was lost to follow-up). Nine pa-tients in the posterior group required contralateralsurgery, whereas 19 did not. The patients who requiredsurgery had mean retroversion of the index shoulderof �18� � 2.4� (95% CI, �19.93� to �16.17�). Theseoperations were performed at a mean of 16.4 � 9months. Eight of the 9 patients had posterior instability,with no identifiable risk factor, cause, or differencenoted in the 1 anterior instability case. The mean gle-noid version of the 9 contralateral shoulderswas �16.63� � 3.00� (95% CI, �19.79� to �13.48�).When the glenoid version of the contralateral shoulderwas compared with the index shoulder, a strong cor-relation of 0.87 was found, with no statistical differencenoted between the 2 sides (P ¼ .11). Injury findingswere similar between both sides, with few shoulderssustaining humeral or glenoid lesions. Thirteen patientsin the anterior group required contralateral surgery,whereas 101 did not. These operations were performedat a mean of 24.4 � 8.4 months. A significantly higherpercentage of patients in the posterior group required

Table 4. Injury Mechanisms for Primary Shoulder

Baseball Basketball Football Snow Sports T

Posteriorn 0 1 11 1% 0 4 39 4

Anteriorn 9 10 26 4% 8 9 23 3

contralateral surgery: 32% versus 11% (P ¼ .01).Retroversion of more than �16� on the index side wasassociated with a higher incidence of posterior contra-lateral injuries (P < .036).Glenoid and humeral bone lesions, diagnosed at the

time of arthroscopy, were available for review in 139 ofthe 143 patients (4 patients had no mention of injury intheir operative report). The posterior group had 6 gle-noid bone lesions and 5 humeral bone lesions. Theanterior group had 32 glenoid bone lesions and 73humeral bone lesions. The anterior group had a statis-tically significantly higher percentage of humeral bonelesions (i.e. Hill-Sachs lesions) than the posterior group:66% versus 18% (P ¼ .0001). When multivariate andunivariate analyses were conducted, 2 statistically sig-nificant relations were found: First, patients with pos-terior instability without a humeral bone lesion showeda significantly higher incidence of contralateral insta-bility (P ¼ .002). Second, patients with a glenoid bonelesion had a higher incidence of a humeral lesion andvice versa (P ¼ .035). No difference was noted in theanterior instability group with relation to glenoid bonelesions, humeral bone lesions, or any other variableincluded.Data on reinjury or failure requiring reoperation were

available for 142 of the 143 patients. Two patients inthe posterior group required reoperation, whereas 14patients in the anterior group required reoperation.This difference was not statistically significant (P ¼ .74).These operations were performed at a mean of 4 � 1months and 46 � 38 months in the posterior group andanterior group, respectively. Surgery was performed incases of either ipsilateral or contralateral instability aftermultiple dislocation events and failure of nonoperativemeasures had occurred. All reinjuries and contralateralinjuries were initially trauma related. No trend wasnoted regarding glenoid version and the need for arepeat operation. These injury patterns are summarizedin Tables 2 and 3.

DiscussionOur study was able to show that patients with pos-

terior instability have a significantly increased amountof glenoid retroversion. A significantly higher percent-age of patients in the posterior instability group sus-tained contralateral injuries requiring surgery: 32%

rauma Water Sports Weight Lifting Wrestling Total

8 2 3 2 2829 7 11 7

55 5 3 3 11548 4 3 3

Table 5. Injury Mechanisms for Contralateral Shoulder

Baseball Basketball Football Snow Sports Trauma Water Sports Weight Lifting Wrestling Total

Posteriorn 0 0 4 0 1 1 3 0 9% 0 0 44 0 11 11 33 0

Anteriorn 1 1 3 1 6 1 0 0 13% 8 8 23 8 46 8 0 0

492 M. B. GOTTSCHALK ET AL.

versus 11%. This was especially true for those withretroversion greater than 16�. Our data also showedthat patients with posterior instability and no humeralbone lesions were more likely to incur contralateralinjuries than those with humeral lesions. This findingcan likely be explained by the fact that a higher-energymechanism was needed to cause the instability event,that is, one would expect to see glenoid or humerallesions in a patient with a higher-energy mechanism.This can be important in the ability to counsel patientsabout the potential for contralateral injuries. Thesefindings are unique when compared with prior studies.Bradley et al.9 reviewed posterior instability in 90

athletes with unidirectional posterior shoulder insta-bility. They showed increased glenoid retroversion of10.7� in those patients with posterior instability incomparison with 10 age-matched controls. Of their 90patients, 48 were contact athletes and had MRI scansavailable for review. The focus of the article by Bradleyet al. was directed toward outcomes and return to playafter arthroscopic posterior repair, which is in contrastto our study.Owens et al.2 prospectively studied 714 young ath-

letes from 2006-2010. Patients underwent stabilitytesting, strength testing, and bilateral noncontrast MRIscans. Glenoid version, height, and depth; rotator in-terval dimensions; and strength measurements were allrecorded. Of the 714 athletes, 46 sustained instabilityevents, of which 7 were posterior. The 7 patients withposterior instability events showed increased retrover-sion of 15.4�, which parallels our study. This studyshowed that increased retroversion is a prospectivepredictor of the direction of instability in young ath-letes. However, the data set did not distinguish whetherthese were contact-/trauma-related incidents.Our data appear to corroborate the 2 aforementioned

studies2,9; however, our study differs in the subset ofpatients, the variables studied, and the type of control

Table 6. Injury Mechanisms for Reinjury Requiring Repeat Surg

Baseball Basketball Football Snow Sports T

Posteriorn 0 0 1 0% 0 0 50 0

Anteriorn 1 0 3 1% 7 0 21 7

group used. We analyzed other variables and theirrelation to glenoid version, such as glenoid bone le-sions, humeral bone lesions, contralateral instability,and reinjury/reoperation, whichdto our knowledgedhas not been previously reported in the literature.Because our control groups did differ, we believed thata more accurate comparison would involve othertraumatic/contact athlete patients sustaining instabilityevents instead of cadaveric or age-matched controls.23

As such, we included over 100 traumatic/contactathlete patients with anterior instability as the controlgroup. This also seemed more logical when trying tocompare concomitant injuries such as glenoid/humerallesions and contralateral injuries and their relation toglenoid version.

LimitationsOur study included just 28 patients with posterior

instability and MRI scans available for review. Howev-er, this number is still larger than the numbers in manyprevious studies, and those studies with larger samplesizes have not limited their study populations topatients with traumatic causes.6,7,24 Although meanpatient follow-up was just less than 1 year, which mayalter some of the results, the study spans a 6-year in-terval, which is an adequate time frame for patients torevisit their physician in the case of a repeat orcontralateral injury.Like with all previous articles studying glenoid

version and other morphologic characteristics, themethod of radiographic measurement is another po-tential limitation. In our study a single reviewer (A.G.)of MRI scans measured glenoid version at 2 differenttime points to help control for intraobserver variabilityand eliminate interobserver variability. Although ourreviewer was not blinded, which certainly weakens themerit of our study, we thought that MRI characteristicscould potentially tip off the reviewer to which direction

ery

rauma Water Sports Weight Lifting Wrestling Total

1 0 0 0 250 0 0 0

7 1 1 0 1450 7 7 0

POSTERIOR SHOULDER INSTABILITY 493

a patient would have instability and therefore negateany blinding or randomization. Our study used themethod of Poon and Ting10 to evaluate glenoid versionbecause it has significant interobserver and intra-observer reliability. Another limitation of our study isthat MRI scans were used to review glenoid versioninstead of CT scans, as originally described by Poon andTing. However, it should be noted that the optimalmethod for glenoid version measurement is unknown,with many previous reports offering seemingly con-tradictory data.6,10-12,14-22,24

ConclusionsOur data show that patients with posterior instability

have a higher incidence of having a retroverted glenoid.Patients with increased retroversion showed increasedposterior contralateral instability. Furthermore, patientswith posterior instability and no humeral bone lesionsmay be more likely to incur contralateral injuries thanthose with humeral lesions. These data suggest thatglenoid version and concomitant injury patterns maybe used to help physicians counsel patients on theirfuture risks of contralateral injury.

References1. Owens BD, Duffey ML, Nelson BJ, DeBerardino TM,

Taylor DC, Mountcastle SB. The incidence and charac-teristics of shoulder instability at the United States MilitaryAcademy. Am J Sports Med 2007;35:1168-1173.

2. Owens BD, Campbell SE, Cameron KL. Risk factors forposterior shoulder instability in young athletes. Am JSports Med 2013;41:2645-2649.

3. Churchill RS, Brems JJ, Kotschi H. Glenoid size, inclina-tion, and version: An anatomic study. J Shoulder ElbowSurg 2001;10:327-332.

4. Iannotti JP, Gabriel JP, Schneck SL, Evans BG, Misra S.The normal glenohumeral relationships. An anatomicalstudy of one hundred and forty shoulders. J Bone JointSurg Am 1992;74:491-500.

5. Lewis GS, Armstrong AD. Glenoid spherical orientationand version. J Shoulder Elbow Surg 2011;20:3-11.

6. Brewer BJ, Wubben RC, Carrera GF. Excessive retrover-sion of the glenoid cavity. A cause of non-traumaticposterior instability of the shoulder. J Bone Joint Surg Am1986;68:724-731.

7. Hurley JA, Anderson TE, Dear W, Andrish JT,Bergfeld JA, Weiker GG. Posterior shoulder instability.Surgical versus conservative results with evaluation ofglenoid version. Am J Sports Med 1992;20:396-400.

8. Wirth MA, Seltzer DG, Rockwood CA Jr. Recurrent pos-terior glenohumeral dislocation associated with increasedretroversion of the glenoid. A case report. Clin Orthop RelatRes 1994;308:98-101.

9. Bradley JP, Baker CL III, Kline AJ, Armfield DR,Chhabra A. Arthroscopic capsulolabral reconstruction for

posterior instability of the shoulder: A prospective study of100 shoulders. Am J Sports Med 2006;34:1061-1071.

10. Poon PC, Ting FS. A 2-dimensional glenoid vault methodfor measuring glenoid version on computed tomography.J Shoulder Elbow Surg 2012;21:329-335.

11. Amadi HO, Banerjee S, Hansen UN, Wallace AL, Bull AM.An optimised method for quantifying glenoid orientation.Int J Shoulder Surg 2008;2:25-29.

12. Bokor DJ, O’Sullivan MD, Hazan GJ. Variability of mea-surement of glenoid version on computed tomographyscan. J Shoulder Elbow Surg 1999;8:595-598.

13. Bryce CD, Davison AC, Lewis GS, Wang L, Flemming DJ,Armstrong AD. Two-dimensional glenoid version mea-surements vary with coronal and sagittal scapular rota-tion. J Bone Joint Surg Am 2010;92:692-699.

14. Budge MD, Lewis GS, Schaefer E, Coquia S, Flemming DJ,Armstrong AD. Comparison of standard two-dimensionaland three-dimensional corrected glenoid version mea-surements. J Shoulder Elbow Surg 2011;20:577-583.

15. Friedman RJ, Hawthorne KB, Genez BM. The use ofcomputerized tomography in the measurement of glenoidversion. J Bone Joint Surg Am 1992;74:1032-1037.

16. Ganapathi A, McCarron JA, Chen X, Iannotti JP. Pre-dicting normal glenoid version from the pathologic scap-ula: A comparison of 4 methods in 2- and 3-dimensionalmodels. J Shoulder Elbow Surg 2011;20:234-244.

17. Kwon YW, Powell KA, Yum JK, Brems JJ, Iannotti JP. Useof three-dimensional computed tomography for theanalysis of the glenoid anatomy. J Shoulder Elbow Surg2005;14:85-90.

18. Moineau G, Levigne C, Boileau P, et al. Three-dimen-sional measurement method of arthritic glenoid cavitymorphology: Feasibility and reproducibility. Orthop Trau-matol Surg Res 2012;98:S139-S145 (suppl).

19. Raymond AC, McCann PA, Sarangi PP. Magnetic reso-nance scanning vs axillary radiography in the assessmentof glenoid version for osteoarthritis. J Shoulder Elbow Surg2013;22:1078-1083.

20. Rouleau DM, Kidder JF, Pons-Villanueva J, Dynamidis S,Defranco M, Walch G. Glenoid version: How to measureit? Validity of different methods in two-dimensionalcomputed tomography scans. J Shoulder Elbow Surg2010;19:1230-1237.

21. Scalise JJ, Codsi MJ, Bryan J, Iannotti JP. The three-dimensional glenoid vault model can estimate normalglenoid version in osteoarthritis. J Shoulder Elbow Surg2008;17:487-491.

22. von Eisenhart-Rothe R, Graichen H, Mayr HO, Jager A,Wiedemann E, Hinterwimmer S. Analyses and compari-son of 2-D vs. 3-D glenoid versions in atraumatic/trau-matic shoulder instabilitydAre 3-D analyses essential?Sportverletz Sportschaden 2009;23:106-111 [in German].

23. Mintzer CM, Waters PM, Brown DJ. Glenoid version inchildren. J Pediatr Orthop 1996;16:563-566.

24. De Wilde LF, Verstraeten T, Speeckaert W, Karelse A.Reliability of the glenoid plane. J Shoulder Elbow Surg2010;19:414-422.