Secondary Logo

Journal Logo

SECTION I: SYMPOSIUM II: Proximal Humerus Symposium

Proximal Humerus Fracture Sequelae: Impact of A New Radiographic Classification on Arthroplasty

Boileau, Pascal*; Trojani, Christophe*; Chuinard, Christopher*; Lehuec, Jean-Charles; Walch, Gilles

Section Editor(s): Iannotti, Joseph P MD, PhD, Guest Editor; Evans, Peter J MD, PhD, FRCSC, Guest Editor

Author Information
Clinical Orthopaedics and Related Research: January 2006 - Volume 442 - Issue - p 121-130
doi: 10.1097/01.blo.0000195679.87258.6e


Sequelae of fractures of the proximal humerus, as first described by Neer,24,25,26 represent some of the most difficult situations to treat in shoulder reconstruction. Historically, a good functional result has been difficult to achieve because of the complexity of the bone loss and deformity. Furthermore, because of the pathoanatomy, shoulder arthroplasty is an unpredictable treatment for sequelae of proximal humerus fractures.2,3,10,17,26,27,35 Because patients usually are younger and more active than patients treated for osteoarthritis (OA), an accurate preoperative prognosis is crucial when deciding on potential treatment.3,8,26,35

There is disappointingly little consensus on treatment in the literature, which contains case series with heterogeneous patient populations and dissimilar anatomic lesions.2,3,8,10,13,14,16,17,19,21,26,28-30,35 Lack of a valid classification for the sequelae of proximal humerus fractures has delayed decision making regarding surgical treatment. If a classification correlated with outcome were developed, treatment reliably should be improved. Patient-related factors that influence outcome could be identified and, if not controlled, at least be recognized and addressed.

In previous studies5,6 we analyzed the results of shoulder arthroplasty in 71 patients with late sequelae from proximal humeral fractures. Furthermore, we described a new surgical classification that has improved our ability to anticipate postoperative arthroplasty results and provided us with improved treatment algorithms. To validate this new surgical classification, we analyzed the results of unconstrained shoulder replacement in a larger series of sequelae of proximal humeral fractures.


We retrospectively reviewed consecutive 203 patients from a number of institutions with 203 sequelae of proximal humeral fractures treated with a nonconstrained shoulder prosthesis and with a minimum of 2 years of followup (range 24-96 months; mean followup, 42 months).

The two lead authors (PB and CT) examined radiographic data, applying the classification scheme, and correlated this data with clinical outcome. Observers were blinded to patient identifiers, but were the treating surgeon for some of the cases. Inclusion criteria were the presence of a preexisting proximal humerus fracture with subsequent arthroplasty and minimum of 24 months of clinical and radiographic followup. The principles of classifying the sequela are simple and were applied to this series.5,6 Two hundred three patients were classified according to our system. With a true anteroposterior (AP) radiograph and an axillary view (or computed tomography [CT] scan) the following questions need to be answered: (1) is there dislocation of the head fragment? Yes = Type 2; (2) is there a nonunion of the surgical neck? Yes = Type 3; (3) can the prosthesis be implanted without doing an osteotomy of the greater tuberosity? No = Type 4; Yes = Type 1.

There were 91 women and 112 men. The average patient age at the time of shoulder arthroplasty was 61 years (range, 19-87 years). The Neer classification of the initial injury pattern was unavailable for 26 patients (25 from our Type 2 sequelae group and one from our Type 3 group).22,23 Regarding the Type 2 sequelae, they were, by definition, missed chronic fracture-dislocations. If the initial diagnosis was missed, it was because no initial xrays were done. There were 35 two-part fractures, 61 three-part fractures, and 81 four-part fractures. One hundred thirty-two (65%) had nonoperative treatment of the initial fracture, and 71 (35%) had an operative procedure to reduce and fix the initial fracture. There were 34 previous surgeries in the Type 1 group (25 open reduction and internal fixation [ORIF], nine percutaneous pinning [PP]), 14 in the Type 2 group (11 ORIF, 3 PP), 12 in the Type 3 group (7 ORIF, 5 PP), and 11 in the Type 4 group (9 ORIF and 2 PP). Average time between injury and arthroplasty was 65 months.

The shoulder replacements were done using the Aequalis prosthesis (Tornier Inc, Houston, TX), a nonconstrained, modular and adaptable prosthesis.4 Senior shoulder surgeons did the arthroplasty in all cases. Patients were collected from 14 centers. The surgical approach was deltopectoral in all but two patients who had a superior transacromial approach. Tenodesis of the long head of biceps was done in 45 patients (22%) if biceps pathology were present. Postoperatively all patients received directed physical therapy.

There were 84 total shoulder arthroplasties (41%) and 119 (59%) hemiarthroplasties; preexisting shoulder pain secondary to osteoarthritis or greater than 45% glenoid cartilage erosion at the time of surgery led to glenoid resurfacing. Osteotomy with complete detachment of the greater tuberosity was performed in all cases of nonunion of the surgical neck (to implant the prosthesis it was necessary to separate the two tuberosities), and in all cases of severe tuberosity malunion (as the distortion of the anatomy was not tolerable).

All the patients were evaluated clinically with the Constant scoring system,9 preoperatively and at the last followup (range, 24-96 months). Clinical records and operative reports were available and reviewed for all the patients included in the study; records included assessments of range of motion and pain level. Preoperative and postoperative standard radiographs were available for 203 patients who were classified according to our system (Fig 1).

Fig 1
Fig 1:
An illustrated diagram shows the surgical classification of sequelae of proximal humeral fractures.

Statistical analysis was done with the student's paired t test for numeric data and with the Pearson and Spearman correlation test for nonparametric data. A p value of less than 0.05 was significant. Statistical analysis was performed centrally by the Service Biostatistique et Informatique Medicale (Lyon, France).


Of the 203 sequelae, there were 137 impacted fractures with humeral head collapse or necrosis (Type 1), 25 locked dislocations or fracture-dislocations (Type 2), 22 nonunions of the surgical neck (Type 3), and 19 severe tuberosity malunions or nonunions (Type 4) (Tables 1, 2). There were 27 two-part fracture, 44 three-part fractures, and 66 four-part fractures that resulted in two different morphologies of Type 1 sequelae: 83 valgus and 54 varus.

Postoperative Mobility after a Nonconstrained Shoulder Prosthesis
Pain and Functional Results (Constant Score) after a Nonconstrained Shoulder Prosthesis

We found a correlation between the final functional results and the following parameters. A greater tuberosity osteotomy jeopardized (p < 0.0005) the final functional results. Initial nonoperative treatment led to better (p < 0.008) final results than initial open reduction and internal fixation. The functional results of shoulder replacement in Type 1 sequelae (cephalic collapse or necrosis) and Type 2 sequelae (locked dislocation or fracture-dislocation) were better (p = < 0.0005) than in Type 3 sequelae (nonunion of the surgical neck) and Type 4 sequelae (severe tuberosity malunion). In Type 1 and 2 sequelae the mean Constant score was higher (67 points) for TSA than hemi (55 points) (p < 0.0005). The functional result was impaired by any complication (p < 0.03) or revision surgery (p < 0.003). We did not find any correlation between the functional results of the entire series and age, gender, delay between the initial fracture and the prosthesis, or long head of biceps tenodesis.

There were 59 complications (27%), including 14 perioperative complications (or in the first postoperative days), and 45 (21%) postoperative complications (Table 3). There were 14 humeral fractures (7%). Complications occurred in 21 (16%) of the Type 1 sequelae of proximal humeral fractures, eight (32%) of Type 2 sequelae, seven (32%) of Type 3 sequelae, and six (32%) of Type 4 sequelae. The rate of complication was lower (p < 0.0005) in Type 1 than the other groups (16% versus 32%). Type 3 sequelae had the worst functional results with non- constrained shoulder arthroplasty in nonunion of the surgical neck. Type 4 sequelae achieved poor results with nonconstrained shoulder arthroplasty in severe tuberosity malunion.

Shoulder Arthroplasty Complications for Sequelae of Proximal Humeral Fractures

Revision surgery was done in 27 patients (12%). There were 15 revisions in Type 1 sequelae (11%), six in Type 2 (24%), three in Type 3 (13%) and three in Type 4 (16%). One hundred seventy-nine (88%) of 203 patients in the series reported being satisfied or very satisfied with the results.


Our current study confirms our previous results with a larger number of patients, clinically validating the utility of our surgical classification.5,6 Identification of the type of sequelae by our classification can allow surgeons to anticipate the prognosis (Fig 1). We found that compression injuries, which maintained continuity between the greater tuberosity and the diaphysis, likely were to result in moderate sequelae, whereas rotational or shearing injuries with displacement of the greater tuberosity led to major sequelae (Fig 2).

Fig 2
Fig 2:
A diagram shows the relationship between the mechanism of initial fracture and the final sequelae.

We note several limitations. Initially, two observers independently evaluated the radiographs, then in concert; while a limitation, the study provides a foundation both to describe fracture sequelae and to compare this to other rating systems. The study did not account for associated injury, and is merely a description of radiographic findings. Soft tissue injury can have a substantial impact on the clinical outcome of the patient. If there is a longstanding nonunion or malunion of the tuberosities, rotator cuff function is likely impaired; prosthetic reconstruction must account for this fact.

Only patients with Types 1 or 2 sequelae benefited from an unconstrained shoulder replacement (Tables 1, 2). At the time of late sequelae, the deformity of the proximal humerus is moderate, with a slight distortion of the normal anatomy and a tuberosity diaphysis continuity, because the initial injury was intracapsular impaction (Fig 2). In Type 1 sequela, it is of paramount to understand that the greater tuberosity is only slightly displaced and that only the cephalic fragment is impacted on the diaphysis and/or collapsed.5,6,18,20 Therefore, it is not necessary to do an osteotomy of the tuberosities to implant the prosthesis; this holds true for Type 2 sequelae.

An unconstrained shoulder arthroplasty gives predictably good results in Types 1 and 2 sequelae as long as the distorted anatomy is accepted (Figures 3, 4). A modular and adaptable shoulder prosthesis is of particular utility in these difficult situations because it allows the surgeon to place the prosthetic head in better proximity to the greater tuberosity, compensating for the distorted anatomy.4,5,6 In sequelae of valgus or varus impacted fractures (Type 1), this technical innovations assists placement of the prosthetic head just above the greater tuberosity, restoring the anatomy with only slight distortion, obviating tuberosity osteotomy. Because only a slight head cut is needed for an anatomic prosthesis, in cases of fracture sequelae, the reamer is passed directly through the head fragment without cutting the head. A small stem cemented in valgus or a stem with a lateral offset is then used. In the case of severe distortion, a shortened stem can be used.

Fig 3A
Fig 3A:
B. The initial valgus impacted four-part fracture resulted in cephalic collapse or necrosis (Type 1 sequela) is shown. (A) There is tuberosity-diaphysis continuity and a slight distortion of the anatomy; therefore, no greater tuberosity osteotomy is needed. (B) Notice that almost no bone cut has been done and that the prosthesis stem has been inserted through the impacted and malunited head fragment. Good functional results such shoulder replacement with a nonconstrained prosthesis are predictable.
Fig 4A
Fig 4A:
B. A locked fracture-dislocation (Type 2 sequela) is shown. (A) A radiograph shows the humeral head is impacted on the posterior glenoid rim, but there is slight distortion of the anatomy and a tuberosity-diaphysis continuity. (B) A radiograph shows that no greater tuberosity has been needed to implant the stem prosthesis. Again, good functional results are predictable.

In locked fracture dislocations (Type 2), we found that total shoulder arthroplasty was preferable if the dislocation had been present for many months because of the frequent severe glenoid erosion. Low to neutral retroversion was chosen for the prosthesis if the dislocation was posterior.12,14,33 Posterior offset of the prosthetic head may be necessary to cover a posteriorly displaced greater tuberosity. Posterior capsulorrhaphy can help residual instability. For severe glenoid deficiency a bone graft may be necessary. Postoperatively, a neutral or external rotation brace is mandatory for 4 to 6 weeks to prevent recurrent posterior instability.31

Conversely, in Type 3 and Type 4 sequelae, it is not possible to implant a nonconstrained prosthesis without doing a greater tuberosity osteotomy because of the severe distortion of the anatomy (Type 4), or because of a surgical-neck nonunion (Type 3). Therefore, conventional arthroplasty produced poor functional results (Figures 5, 6). Types 3 and 4 sequelae are secondary to displaced and disimpacted extracapsular fractures. At the time of the initial fracture, the greater tuberosity and the humeral head are displaced from the diaphysis and, if left untreated, heal in their displaced location leading to severe deformity or a surgical neck nonunion of the proximal humerus (Fig 2). Therefore, greater tuberosity osteotomy typically is needed to implant an unconstrained prosthesis. As a result, Types 3 and 4 sequelae are relative contraindications to unconstrained prosthesis and other surgical options should be considered. Furthermore, tuberosity osteotomy degrades functional results because of difficulty with healing and impairment of cuff function.

Fig 5A
Fig 5A:
C. Failed treatment with nonunion of the surgical neck (Type 3 sequela) led to hemiarthroplasty (A). This patient was treated with an unconstrained shoulder arthroplasty. (B) Poor, unpredictable results are shown (C) because of persistent non- union, malunion, or osteolysis after a greater tuberosity osteotomy.
Fig 6A
Fig 6A:
C. A four-part fracture with severe tuberosity malunion (A) was treated with hemiarthroplasty (B) and resulted in severely limited forward flexion (C). In Type 4 sequela (disimpacted sequela with severe malunion of the tuberosities), it is impossible to implant a shoulder prosthesis without doing a greater tuberosity osteotomy. Therefore, the results of a nonconstrained shoulder arthroplasty are poor.

Treatment of nonunion of the surgical neck (Type 3) is difficult, and the surgeon should consider options other than an unconstrained prosthesis. Filling of the medullary canal by the prosthesis often prevents union of the tuber- osities to the humeral shaft.1,5,6,11,15,27,32 Therefore, a Type 3 sequela is a relative contraindication to a noncon- strained shoulder prosthesis. Two surgical alternatives are possible. If there is minimal humeral head cavitation and/or osteoarthritis, then surgical treatment with an intramedullary bone peg graft and internal fixation is indicated (Fig 7). The head fragment is usually alive, and our experience with the peg bone graft has been very successful-even with a small head fragment.34 For severe humeral head cavitation and/or osteoarthritis, we prefer a low profile prosthesis with bone graft (Aequalis-Fracture Prosthesis Tornier Inc, Houston, TX). This device improves the chances of healing by allowing the placement of copious bone graft between the epiphysis and the diaphysis. Union of the tuberosities to the shaft should be the primary goal (Fig 8).

Fig 7A
Fig 7A:
C. (A) A radiograph shows the head fragment is small but still alive (Type 3 sequela). (B) An illustration shows how an intramedullary peg bone graft is taken from the iliac crest. (C) A radiograph shows bone healing after osteosynthesis with a bladeplate and an intramedullary peg bone graft. This technique has yielded excellent functional results 5 years postoperatively.
Fig 8
Fig 8:
A radiograph shows that a low profile fracture- prosthesis allows ample bone graft placement between tuberosities and shaft (Aequalis Fracture-prosthesis, Tornier Inc, Houston, TX).

The results of a nonconstrained shoulder arthroplasty are predictably poor in Type 4 sequela (unimpacted with severe tuberosity malunion). Therefore, a Type 4 sequela is also a contraindication to a nonconstrained shoulder prosthesis. An alternative is a semiconstrained reverse ball and socket prosthesis (Delta prosthesis, DePuy, Warsaw, IN or Aequalis-Reversed Prosthesis, Tornier Inc, Houston, TX). Early results (pain relief and function) of this prosthesis are better than those achieved with an unconstrained prosthesis and concomitant greater tuberosity osteotomy (Fig 9).7 Reverse prosthetic replacement can provide a good functional result in the face of a deficient rotator cuff-if the deltoid muscle is working.

Fig 9A
Fig 9A:
B. (A) A Type 4 sequela (severe distortion of the anatomy with severe tuberosity malunion and nonunion after previous pinning) is not a candidate for a conventional nonconstrained prosthesis. (B) A reverse prosthesis (Aequalis reverse prosthesis, Tornier, Inc., Houston, TX) is the prosthesis of choice to provide good functional results.

We show a practical rationale for the classification of old fractures of the proximal humerus for shoulder arthroplasty. When deciding the treatment of a proximal humeral fracture sequela one must consider: (1) the distortion of the proximal humeral anatomy; (2) the tuberosity-diaphysis continuity, and (3) the need of a greater tuberosity osteotomy (Table 4). If the distortion of the anatomy is moderate and there is tuberosity-diaphysis continuity (Types 1 and 2 sequelae), then noncontrained shoulder prosthesis can be done without performing a greater tuberosity osteotomy, and good functional results can be expected. The prosthesis and technique should be modified to accommodate the distorted anatomy.

Surgical Indications in Proximal Humeral Fracture Sequelae

If there is tuberosity-diaphysis discontinuity (Type 3 sequela) and/or a severe distortion of the anatomy (Type 4 sequela), a greater tuberosity osteotomy is needed, with predictably poor functional results of a nonconstrained shoulder replacement. Therefore, we no longer use an unconstrained prosthesis in Type 3 and Type 4 fracture sequelae. For Type 3 sequelae, we recommend intramedullary peg bone graft with osteosynthesis. A low profile fracture prosthesis can also be indicated if the head cavitation is severe, or if there is osteoarthritis; finally, this allows bone grafting of the surgical neck nonunion. We recommend a reverse prosthesis for Type 4 sequela (Fig 10).

Fig 10
Fig 10:
Updated surgical indications in proximal humeral fracture sequelae: nonconstrained prosthesis in type 1 and 2; ORIF and peg bone graft in type 3 and reverse prosthesis in type 4.


The authors thank Drs. Adrian, Allieu, Armengol, Basso, Brunner, Burger, Candau, De Beer, Deprey, Duparc, Fama, Favard, Franceschi, Gazielly, Gerber, Gohlke, Goutallier, Huguet, Kelberine, Kempf, Laurent, Le Huec, Léger, Levigne, Loehr, Loew, Matsoukis, Mestdagh, Molé, Mora, Nové-Josserand, Oudet, Pfahler, Scheffer, Van der Maren, Versier, Wiedeman, and Willmens for providing some of the patients for the study. The authors also thank Professor. P. Adeleine, PhD and M. Canova, BA for the statistical analysis.


1. Antuna SA, Sperling JW, Sanchez-Sotelo J, Cofield RH. Shoulder arthroplasty for proximal humeral nonunions. J Shoulder Elbow Surg. 2002;11:114-121.
2. Antuna SA, Sperling JW, Sanchez-Sotelo J, Cofield RH. Shoulder arthroplasty for proximal humeral malunions: Long-term results. J Shoulder Elbow Surg. 2002;11:122-129.
3. Beredjiklian Pk. Iannotti JP, Norris TR, Williams GR. Operative treatment of malunion a fracture of the proximal aspect of the humerus. J Bone Joint Surg. 1998;80A:1484-1497.
4. Boileau P, Walch G. The three-dimensional geometry of the proximal humerus: Implications for surgical technique and prosthetic design. J Bone Joint Surg. 1997;79B:857-865.
5. Boileau P, Walch G, Trojani C, Sinnerton R, Romeo AA, Veneau B. Sequelae of Fractures of the Proximal Humerus: Surgical Classification and Limits of Shoulder Arthroplasty. In: Walch G, Boileau P (eds). Shoulder Arthroplasty. Berlin, Germany: Springer-Verlag 349-58, 1999.
6. Boileau P, Trojani C, Walch G, Krishnan S, Romeo A, Sinnerton R. Shoulder arthroplasty for the treatment of the sequelae of fractures of the proximal humerus. J Shoulder Elbow Surg. 2001;10:299-308.
7. Boileau P, Watkinson D. A Hatzidakis,Balg F. The Grammon reverse prosthesis: design, rational, and biomechanics. J Shoulder Elbow Surg. 2004;14:147-161s.
8. Cofield RH. Shoulder replacement: prognosis related to diagnosis. In: Kôlbel R, Helbig B, Blauth W, eds. Shoulder Replacement: Prognosis Related to Diagnosis. Shoulder Replacement. Berlin, Germany: Springer-Verlag 157-161, 1987.
9. Constant CR,Murley AHG. A Clinical Method of functional assessment of the Shoulder. Clin Orthop Relat Res. 1987;214:160-164.
10. Dines DM, Klarren RF, Altcheck DW, Moeckel B. Posttraumatic changes of the proximal humerus: Malunion, nonunion, and osteonecrosis. Treatment with modular hemiarthroplasty or total shoulder arthroplasty. J Shoulder Elbow Surg. 1993;2, 1 :11-21.
11. Duralde XA, Flatow EL, Pollock RG, Nicholson GP, Self EB, Bigliani LU. Operative treatment of nonunions of the surgical neck of the humerus. J Shoulder Elbow Surg. 1996;5, 3 :169-180.
12. Flatow EL, Neer CS II. Chronic anterior dislocation of the shoulder. J Shoulder Elbow Surg. 1993;2, 1 :2-10.
13. Frich LH, Sjoberg JO, Sneppen O. Shoulder arthroplasty in complex acute and chronic proximal humerus fractures. Orthopaedics. 1991;14:949-954.
14. Hawkins RJ, Neer CS II. Locked posterior dislocation of the shoulder. J Bone Joint Surg. 1987;69A:9-18.
15. Healy WL, Jupiter JB, Kristiansen TK, White RR. Nonunion of the proximal humerus. In: Post M, Morrey BF, Hawkins RJ, eds. Non- union of the proximal humerus. Surgery of the Shoulder. St. Louis: Mosby-Year Book 59, 1990.
16. Huten D, Duparc J. L'arthroplastie prothétique dans les traumtismes complexes récents et anciens de l'épaule. Rev Chir Orthop. 1986;72:517-529.
17. Ianotti JP, Sidor ML. Malunions of the proximal humerus. In: Warner JP, Ianotti JP, Gerber C, eds. Complex and Revision Problems in Shoulder Surgery, Philadelphia: Lippincott-Raven 245-264, 1997.
18. Jakob RP, Miniaci A, Anson PS. Four part valgus impacted fractures of the proximal humerus. J Bone Joint Surg. 1991;73B:295-298.
19. Keene JS, Huizenga RE, Engber WD, Rogers SC. Proximal humeral fractures: a correlation of residual deformity with long term function. Orthopedics. 1983;6:173.
20. Kyle RF, Conner TN. Post-traumatic avascular necrosis of the humeral head in displaced proximal humeral fractures. J Trauma. 1981;21:788-791.
21. Lee CK, Hansen HR. Post-traumatic avascular necrosis of the humeral head in displaced proximal humeral fractures. J Trauma. 1981;21:788-791.
22. Neer CS II. Displaced proximal humeral fractures part 1: classification and evaluation. J Bone Joint Surg. 1970;52A:1077-1089.
23. Neer CS II. Displaced proximal humeral fractures part 2: treatment of three and four part displacement. J Bone Joint Surg. 1970;52A: 1090-1103.
24. Neer CS II, Watson KC, Stanton FJ. Recent experience in total shoulder replacement. J Bone Joint Surg. 1982;64A:319-337.
25. Neer CS II. Nonunion of the surgical neck of the humerus. Orthop Trans. 1983;3:389.
26. Neer CS II. Old trauma in glenohumeral arthroplasty In: Shoulder Reconstruction. Ed 3. Philadelphia: Saunders 222-234, 1990.
27. Norris TR, Turner JA, Bovill D. Nonunion of the upper humerus: an analysis of the etiology and treatment in 28 cases. In: Post M, Morrey BF, Hawkins RJ (eds). Surgery of the Shoulder. St. Louis: Mosby-Year Book 63, 1990.
28. Norris TR, Turner JA. Late prosthetic shoulder arthroplasty for displaced proximal humerus fractures. J Shoulder Elbow Surg. 1995;4:271-280.
29. Pritchett JW. Prosthetic replacement for chronic unreduced dislocations of the shoulder. Clin Orthop Relat Res. 1987;216:89-93.
30. Razemon JP, Baux S. Les séquelles des fractures de l'extrémité supérieure de l'humérus. Rev Chir Orthop. 1969;55:479-490.
31. Rowe CR, Zarins B. Chronic unreduced dislocations of the Shoulder. J Bone Joint Surg. 1982;64A:494-505.
32. Scheck M. Surgical treatment of nonunions of the surgical neck of the humerus. Clin Orthop Relat Res. 1982;167:255-259.
33. Walch G, Boileau P, Martin B, Dejour H. Unreduced posterior dislocations and fracture-dislocations of the shoulder. A review of 30 cases. J Orthop Surg. 1990;4:461-472.
34. Walch G, Badet R, Nove-Josserand L, Levigne C. Nonunions of the surgical neck of the humerus: surgical treatment with an intramedullary bone peg, internal fixation and cancellous bone grafting. J Shoulder Elbow Surg. 1996;5:161-168.
35. Watson KC. Indications and considerations of shoulder replacement in post-traumatic conditions. In: Köbel, Helbig and Blauth, eds. Shoulder Replacement. Berlin: Springer-Verlag129-133, 1987.
© 2006 Lippincott Williams & Wilkins, Inc.