FIXATION OF FRACTURES
OF THE PROXIMAL HUMERUS
USING THE BILBOQUET DEVICE
Surgical technique and indications
Hôpital Européen Georges Pompidou - Paris, France
Fracture fixation with the Bilboquet device provides a solution to the majority of problems encountered in complex fractures of the proximal humerus: the anatomy can be reconstructed, implant impingement on the surrounding soft tissues is prevented, and early mobilization of the limb is facilitated.
The device may also be readily converted into a hemiarthroplasty, should the humeral head undergo avascular necrosis.
The only disadvantage is that the implant is fitted into the cancellous bone of the humeral head, and cannot, therefore, be removed at a later stage.
The Bilboquet technique for the fixation of complex fractures of the proximal humerus was evolved over a period of 10 years. Gathering a sufficiently large patient material to allow evaluation took time and patience, since these fracture patterns are not very common. Jean Grimberg wrote a thesis on the subject, which provided the basis of a paper published in the Revue de Chirurgie Orthopédique, in 1996.1 More recently, a paper reporting 26 cases was published in the Journal of Shoulder and Elbow Surgery.2 In the light of the results obtained, this technique may now be recommended for use in appropriately selected patients, in whom it simplifies the management of fractures of the proximal humerus and obviates the need for shoulder replacement.
Proximal humeral fractures (PHF) have two important features: the metaphysis is comminuted, and there is a risk of avascular necrosis of the humeral head.
These fractures commonly occur in elderly subjects with osteoporotic bone stock. The metaphysis is broken into several fragments. This means that, in the majority of cases, there will not be a sound metaphyseal portion to provide support for the fixation construct; in particular, there will not be sound support for the humeral head fragment.
One option would be to forego anatomical reconstruction, and to fixate the fragments in their impacted position in the metaphysis. However, this would result in a deficit in humeral height, and would impair the functional outcome. If the head is to be restored to its correct level, a distraction construct would need to be employed; the outcome would be unpredictable. The humeral head is a one-third of a sphere filled with cancellous bone, which offers little purchase to conventional fixation hardware.
Fig. 1 Fig. 2 Fig. 3
The unpredictable nature and poor outcomes of internal fixation had prompted C.S. Neer to counsel prosthetic replacement.3,4 However, the literature published at the time of Neer's original papers shows that the Neer technique was not solely designed to replace the humeral head: the Neer device also served to restore the fractured humerus to its correct height, and to replace the tuberosities, whose importance for the reconstruction Neer was among the first to stress.
Restoration of humeral height and repositioning of the tuberosities are, indeed, imperative; and shoulder surgeons have followed Neer's precepts in this respect. Unfortunately, the contused and osteoporotic tuberosities in elderly patients will heal poorly if they are held tight against a piece of metal. There is general agreement that the outcome of humeral hemiarthroplasty in trauma patients is disappointing.5,6,7
This has led to a more conservative approach, which has attempted to combine the benefits of internal fixation with those of Neer's technique.
The Bilboquet device was developed with a view to resolving the mechanical problem, inherent in the internal fixation of PHF, of how to distract the head without allowing it to tilt.
This problem is difficult to resolve if the only point of fixation is in the centre of the humeral head. However, if the insertion of the hardware is spread around the circumference of the fragment, the loads involved will be better distributed over the surface of the fragment, and stability will be enhanced (Figs. 1-3).
Avascular necrosis is a frequent complication; its clinical repercussions vary greatly.
If necrosis occurs in a trauma patient whose tuberosities have been badly repositioned, the poor clinical outcome will be wrongly attributed to necrosis. However, isolated necrosis in a well-reconstructed proximal humerus may go entirely unnoticed. There is a major difference between necrosis of the humeral head and necrosis of the femoral head. In the first instance, this is due to the different loading patterns at these sites. In the second instance, the difference may be due to easier bony remodelling at the humeral site, which does not have a neck.
There are numerous reports to show that-for whatever reason-partial avascular necrosis is well tolerated in trauma cases.
A poor functional outcome is chiefly due to failure of the tuberosities.
The Bilboquet is designed in such a way as to permit conversion from an internal fixation construct to a hemiarthroplasty, by replacing the head staple with a prosthetic humeral head. Our original fears that conversion would have to be resorted to in a large number of cases have proved largely unfounded, since necrosis has been seen to be well tolerated: the tuberosities tend to heal well, and the clinical outcome (ROM, freedom from pain) will be good. Preservation of the humeral head means that the tuberosities are repositioned in contact with bone, which provides a much more favourable environment than does metal.
The patient is placed in a semisitting position, with the affected shoulder slightly beyond the edge of the operating table.
We do not use a pad under the shoulder or the thoracic spine.
The upper limb must be draped entirely free, to allow mobilization as required during the procedure; a wide arm support beneath the elbow might obstruct the surgeon.
The patient's head should be immobilized with one or two strips of adhesive tape, so as to keep it steady when traction is exerted on the arm. There are, however, certain precautions that must imperatively be observed:
- The cervical spine must not be hyperextended. Hyperextension would be particularly hazardous to the cervical cord in patients with cervical spondylarthrosis. All that is required in order to guard against hyperextension is to place a sufficiently large number of cushions behind the patient's head to ensure that the head is in a natural position.
- The patient's head must not be turned towards the unaffected side, since doing so would cause unnecessary tension in the brachial plexus.
Draping is performed in customary fashion. An antibacterial adhesive drape is applied over the surgical site.
The fracture site is approached via a deltopectoral incision. Once the superficial tissues have been divided, the incision may be deepened by blunt dissection of the deltopectoral groove. The fracture haematoma will have disrupted the tissues, and, in many cases, finger dissection will readily create a deltopectoral space (Fig. 4).
The haematoma must be evacuated, and the first stage of fracture site exploration is done with the probing finger (Fig. 5). This is a non-traumatic means of checking the position, size, and condition of the fracture fragments. To the surgeon, who will have analyzed the patient's radiographs preoperatively, this digital exploration yields more information than does a wide (and tissue-damaging) dissection of the fracture site. Once the exploration is completed, the gloves should be checked for cuts and punctures.
A self-retaining rake retractor is inserted, to allow visual inspection of the fracture site (Fig. 6) and, if required, resection of part of the subdeltoid bursa swollen by the haematoma. In some cases, a small double-angled retractor may be placed above the coracoid process, to provide improved exposure.
Fig. 5 Fig. 6 Four-part fracture. The fragment bearing the lesser tuberosity is in the foreground.
A vertical deltoid-splitting incision, as used in rotator cuff surgery, may be employed, provided that the lower fracture line is just below the tuberosities. In this approach, the axillary nerve will limit an extension of the incision downwards, and metaphyseal/diaphyseal fractures cannot be inspected. The deltoid-splitting approach does, however, permit improved inspection of any posterior displacement of the greater tuberosity.
Three- or four-part fractures
Implant insertion is prepared in three stages, which may overlap:
- exposure and preparation of the humeral shaft;
- exposure of the fracture surface of the head fragment;
- exposure and preparation of the tuberosity fragments.
The general principle is to work within the fracture site, so as to cause as little additional soft-tissue trauma as possible.
Case No. 1
Age 78 years. Four-part fracture.
(a) Preoperative radiograph
(b) Immediate postoperative radiograph
(c) Check radiograph at 3 months, showing healing
(d) Good clinical outcome at 2 years, patient pain-free, despite radiographic evidence of partial necrosis of humeral head. Note excellent healing of greater tuberosity.
Bilboquet Mark I. Stem and staple made of titanium. Staple available in three sizes. New version of the Bilboquet. Spigot identical to that on the Solar humeral prosthesis (Stryker Howmedica)
Exposure and preparation of the humeral shaft
In a few cases, the shaft will be found to be badly displaced, with its proximal end sitting anteriorly. In such a case, the medullary canal will be readily accessible.
More often than not, however, the shaft will be found to be sitting medially, with its proximal end covered by the impacted head fragment. In such cases, the fragments will need to be moved, so as to provide access to the medullary canal.
First, the tuberosities will have to be held out of the way (Fig. 7). Next, the head will need to be prised off gently with an elevator, and lifted, applying thumb pressure to its cancellous surface (Fig. 8). This manoeuvre is intended to lift the head off the shaft, and to place it against the glenoid, without damaging the cancellous bone or exerting undue traction on any vessels supplying the head that may have survived the fracture.
Fig. 7 Fig. 8
Once the head has been lifted out of the way, the shaft is mobilized to provide access for the insertion of the implant into the medullary canal. If mobilization proves difficult, it may be facilitated by dividing the proximal portion of the pectoralis major insertion on the humerus. A bone hook may be have to be used (Fig. 8). The general principle is to preserve as much as possible of the humeral head and its attachments, and to confine mobilization to the humeral shaft.
In some cases, the humeral head can be lifted only if the tuberosities are first held out of the way; in the majority of four-part patterns, however, the greater tuberosity will not be in sight.
Where the fragments are holding together well, there is no need for the use of the Bilboquet. Sometimes, lifting the head will suffice to reduce the tuberosities. (Once lifted, the head may be retained in this position with a wedge of bone substitute, or by resting it on the reduced tuberosities.) Wiring the tuberosities to the shaft will then effect a minimal-fixation construct, which allows early rehabilitation with pendulum exercises, and should provide an excellent outcome.
Preparation of the shaft involves
1- Manual reaming (Fig. 9);
Fig. 9 Fig. 10
2- Drilling two holes, 2-3 cm below the fracture site, through which a stout suture will be passed to re-attach the greater tuberosity (Figs. 9, 10);
3- Determining the optimal level of the humeral implant.
Usually, the level is referenced on the medial border of the humeral shaft. A line drawn through the neck base of the definitive stem should form a tangent to the medial cortex of the shaft (Fig. 11). Laterally, this may look like an excessively proud implant; however, it should be borne in mind that the fracture will have been at the expense of the metaphysis, and that the lifting of the head will have left a lateral bone defect (Fig. 12).
Fig. 11 Implant trial at correct level - note swab used as wedge. Fig. 12
View in internal rotation
Roughly speaking, the proximal end of the humeral shaft is flared, with the medial aspect more or less in continuity with the head fragment, while the lateral aspect has been deprived of the tuberosities, which will have been displaced by the fracture. With the tuberosities not in their usual position, it is not surprising to see that the lateral aspect of a prosthetic or a Bilboquet stem will be exposed over a distance of several centimetres. This lateral defect should be fully covered once the tuberosities have been reduced onto the shaft.
In very rare cases, the medial aspect of the shaft will be involved in the fracture pattern, and, hence, not available as a reliable landmark (Case No. 4).
Since the range of stem sizes is limited, it may be necessary to wedge the trial stem into the medullary cavity, using a gauze swab, in order to obtain implant seating at the correct level.
Case n° 4 Age 71 years. Badly comminuted metaphyseal/diaphyseal fracture pattern. The only problem consists in controlling the level of the humeral implant. Once the stem has been reduced into the staple, the fracture site is aligned, leaving only the metaphyseal/diaphyseal fragments to be reattached. View at end of procedure. Postoperative radiograph. Check radiograph at 6 months. The clinical outcome was good.
Exposure of the fracture surface of the head fragment
Lifting the humeral head serves to expose the cancellous bone of the head. Inspection and palpation have to be used in order accurately to define the edges and determine the orientation of the head.
The staple is placed against the cancellous surface in such a way as to obtain maximum coverage. The staple has a central Morse taper, which inserts into the centre of the humeral head, while the prongs around the circumference insert into the periphery of the fracture surface. Choosing too small a size would create the following problems:
- the staple would be off-centre, and would, therefore, be difficult to align; and
- peripheral stabilization of the humeral head would be lost.
A correctly sized staple, on the other hand, has two advantages: staple alignment will be straightforward, since the staple base will become the base of the head hemisphere; and, with the prongs biting into the periphery of the fracture surface, there will be little risk of the head tilting.
If the staple size is unduly large, the prongs would be protruding beyond the edge of the head.
In practice, the more severe the fracture, the easier the insertion of the staple will be: in a head that has come off most of its attachments, the size and edges of the fracture surface will be easier to define, and staple size selection as well as the insertion of the staple will be greatly facilitated.
Fig. 13 Fig. 14 : Not correct Fig. 15 : correct
Whether the humeral head should be retained or sacrificed will depend on two factors:
- Patient age. The younger subject, the more reason there would be to preserve the head. Even where the head has been totally deprived of its attachments, using the head fragment as an osteochondral autograft would seem preferable to the use of a hemiarthroplasty. In these younger patients, such grafts stand a good chance of taking, and any necrosis that may occur may be only partial and well-tolerated.
- Displacement. Traumatic two-part fracture-dislocations, seen mainly in anterior dislocation, are, in theory, not at risk of necrosis, since the head will still have a blood supply via the capsular insertions below. We have also found that many heads which appeared at first sight to be completely detached would resist gentle efforts at extraction, since, in their lower part, they were still soundly attached by a sheet of capsular tissue.13 In many of these cases, retaining the head was followed by healing without avascular necrosis.
Total detachment of the head fragment in the elderly is almost invariably followed by complete avascular necrosis; fixation is, therefore, of no benefit. These cases constitute an absolute indication for hemiarthroplasty.
Exposure and preparation of the tuberosity fragments
This stage of the procedure is of fundamental importance. It is performed in the same way as for the insertion of a humeral prosthesis. The fragments must be carefully identified. The greater tuberosity is often displaced far posteriorly and medially, and must be brought back gradually, using forceps or suture traction (Figs. 16, 17).
Axial view of displaced tuberosity fragments
Traction applied to posterior tuberosity fragment, to permit suturing
Fracture fixation (Case No. 3)
(a) Age 51 years. Displaced four-part fracture. Broken line indicates position of tuberosities.
(b) Deltopectoral approach. The partially attached tuberosities are in the foreground.
(c) Fracture site distraction reveals humeral head posteriorly, and the two tuberosity fragments superiorly and partially attached on either side of the biceps tendon.
(d) Placement of staple over the cancellous bone of the humeral head fracture surface.
(e) Impaction of staple into the cancellous bone of the humeral head, using the staple holder. Following impaction, the green part of the handle is turned to disengage the holder.
(f) Using a small curette to remove surplus cancellous bone from inside the female Morse taper of the staple.
(g) Preparation of humeral shaft.
(h) Reduction of trial stem into the head staple.
(i) Cementing of definitive component (in this case, a SolarÒ stem). Lower photograph shows use of swab to wedge trial stem at correct level. The implant will be at the appropriate level if the line of the collar continues as a tangent to the proximal edge of the medial cortex. There should be as little metal as possible in the metaphysis.
(j) View at end of procedure, and immediate postoperative radiograph.
(k) Excellent clinical and radiographic result at 6 months.
- The staple is placed in the staple holder, and impacted into the cancellous bone of the humeral head. During impaction, the staple holder may be temporarily removed, to allow removal of the cancellous bone from inside the Morse taper, with a small curette. If left inside the taper, the bone may interfere with the advance of the staple. In the absence of a staple holder, a large-diameter graft pusher may be used. Where a graft holder is being used, care should be taken to ensure that only the green part of the handle is turned to remove the holder from the staple; turning the red part of the handle would twist the staple in its cancellous bed.
- A humeral trial of the required size is placed inside the shaft, and a trial reduction is performed. If the spigot on the stem engages the female taper inside the staple very easily, the stem is not sufficiently proud of the shaft, and the construct will lack height. If, on the other hand, the stem is unduly proud, reduction may be very difficult, or overpressure in the humeral head may force the staple out.
- As regards retroversion, I tend to use ca. 30° of retroversion, referenced on the bicondylar line at the elbow. As a general rule, one may go by the position of the spigot on the stem, which, at rest (i.e. in physiological internal rotation) should be pointing towards the head staple.
- Engagement is brought about by exerting downward traction on the forearm flexed at 90°, and holding the stem at its upper end to guide it. At the same time, the assistant exposes the fracture surface of the humeral head with an instrument such as a narrow elevator, to facilitate the engagement of the spigot in the internal taper of the staple (Fig. 18). A certain amount of traction is required, in order to restore the correct anatomical height.
Prior to this manoeuvre, the anaesthetist should be asked to check that the patient's head is well immobilized; during the manoeuvre, a check should be made to ensure that the trial stem is firmly seated in the humeral shaft.
Once the correct height has been established, the trial stem is removed by pulling on its wing, while firmly pushing the staple against its cancellous bed, with an elevator. The definitive stem is inserted. Prior to its insertion, a nonresorbable suture is passed through the holes in the shaft, for subsequent use in the attachment of the tuberosities.
If the stem is perfectly seated in the shaft, at the required level (which will very rarely be the case), cementing may be dispensed with. However, since the stem is subject to major rotatory stresses, cement will, as a rule, need to be used. Cementing is done in customary fashion (drying of the medullary cavity, suction drain). It should, however, be noted that cement is used purely to fix the stem in the diaphysis; consequently, there is no need for using cement in the metaphysis. While the cement is setting, care should be taken to prevent the implant losing height or changing its degree of retroversion.
One very useful instrument for the engagement of the definitive stem in the staple is a good old-fashioned towel clamp. The clamp is inserted into one of the holes in the lateral wing of the stem, and allows the spigot to be aligned with the staple during reduction.
- If the stem is at the correct level, reduction will stabilize the fracture site even in the absence of the tuberosities, and passive movements imparted to the arm will be transmitted to the humeral head. Also, the tuberosities will have been restored to their correct anatomical position, without any overriding or gaps.
- The tuberosities are replaced in their correct anatomical positions, and the sutures are tied. We commonly use two horizontal ties, plus a vertical figure-of-eight tie passed through the supraspinalis at the top, and through the holes in the shaft below (Fig. 19).
Since the head has been retained, the tuberosities will be in contact with native bone, which probably accounts for the fact that tuberosity non-union is rarely observed following this procedure.
At the end of the procedure, the shoulder should move freely through the entire ROM. It should be borne in mind that the usual form of postoperative immobilization is a simple Mayo Clinic bandage. If the internal rotation, with the elbow against the body, imposed by this bandage causes too much pull on the tuberosities, immobilization should be in slight internal rotation, with an abduction pad placed under the arm. It may be safer to apply a full-scale swathe.
Displaced two-part (surgical-neck) fractures are, by and large, very amenable to internal fixation with conventional hardware. Using a permanent implant for their repair would not, therefore, be desirable. However, there are situations where a permanent device may be indicated. Thus, in elderly and osteoporotic subjects, conventional fixation hardware is notoriously difficult to insert stably in the head fragment. In such cases, the Bilboquet provides a simple and effective solution to the problem of fracture site stabilization. Similarly, where the fracture involves the metaphysis and the diaphysis (Case No. 4), the reconstruction achieved with the Bilboquet is far superior to what can be expected from fixation with conventional hardware.
In essence, the technique for the management of two-part fractures is the same as that employed for the fixation of three- or four-part patterns; however, there are some special features to be borne in mind.
Staple insertion is more difficult than with the multiple-fragment patterns. In the two-part pattern, the head will be reasonably intact, which makes it more difficult to align and insert the staple. The edges of the head must be carefully defined, to avoid two particular pitfalls -
- axial malalignment: as a rule, the staple is inserted too far anteriorly (the extreme being insertion into the cancellous bone of the lesser tuberosity);
- coronal malalignment: this involves the risk of excessively steep positioning of the staple (the extreme being insertion in the cancellous bone of the greater tuberosity).
As noted above, using the largest staple size that can be accommodated on the fracture surface is the best safeguard against malalignment. Sometimes, an intact humeral head will not allow the staple to be inserted, in which case a small amount of bone will need to be removed in the lateral metaphysis, to facilitate insertion.
The immediate stability of the construct is often very satisfactory, and the surgeon may be tempted to dispense with the tension banding of the tuberosities. This would be very misguided: in order to prevent the head tilting into varus, the tuberosities must be laterally banded to the shaft, to complete the construct.
The immobilization to be used will be a function of the general circumstances and of the soundness of the construct. As a general rule, a Mayo Clinic bandage is worn for the first few days, to be replaced by a simple sling. Physiotherapy should be started early, but very gently. We initiate passive mobilization at 48 hours; this programme is continued at least until Week 3. Thereafter, active exercises are gradually introduced. The minimum duration of rehabilitation is 3 months; full functional recovery frequently takes 6 months.
- The chief complication following internal fixation of complex PHF is avascular necrosis of the humeral head. This necrosis may be partial, and well tolerated. In symptomatic patients, the Bilboquet device may be readily converted to a hemiarthroplasty. All that is required is the removal of the head staple, and the fitting of a humeral head replacement onto the stem spigot.
- Protrusion of staple prongs. In our experience, this complication does not usually produce any symptoms. In one case, with the humeral head in varus, a superiorly protruding prong had caused rotator cuff ulceration and inflammatory-type pain. The patient had to be reoperated on, with conversion to a hemiarthroplasty.
- Malalignment of the head in varus. This complication is observed where fixation is incomplete or insufficiently strong. It is all the more likely to occur where the staple is of insufficient size, and where the tension banding is ineffective. In our experience, it tends to cause restriction of the ROM rather than non-union or necrosis.
- Dislocation of the construct. In theory, this is a risk inherent in constructs of insufficient height. In practice, we have encountered this complication only once, in an elderly lady who fell out of bed postoperatively. The implant was reduced, healing was obtained, and there have been no further problems during the 6-year follow-up to date.
- Infection. We have not, to date, had a single case of infection. The good regional blood supply, the use of cement containing antibiotics, the prophylactic administration of antibiotics, and the fact that the implant is "shielded" by bone from the surrounding tissues and haematoma, may account for this absence of infection. Also, the number of patients managed to date with this device has been comparatively small. Obviously, there will be a case of infection, sooner or later. In early postoperative cases, immediate reoperation, with lavage, drainage, and antibiotics may be all that is required. Infection within the medullary cavity would necessitate the removal of the implant and, consequently, of the humeral head.
The Bilboquet constitutes an alternative to hemiarthroplasty, since it allows the sound internal fixation of complex fractures that could not be managed with conventional fixation techniques. All fracture patterns may be repaired using this device; whether it should be used is a function of the risk of avascular necrosis, which must be assessed at surgery in each individual case. In an elderly patient, the risk may be high, in which case it would be better to perform a hemiarthroplasty.
The Bilboquet is also suitable for the fixation of displaced PHF in elderly patients with osteoporotic bone stock. Providing that the patient is not a poor anaesthetic risk, these fractures (which are common in the elderly) may be repaired regardless of patient age, and with a prospect of a good functional outcome.
Specifically, the Bilboquet is indicated in
- Complex fractures of the proximal humerus. In our practice, "complex fractures" is defined as displaced three- and four-part fractures (based on the classification by Neer), or Type II and III head and tuberosity fractures (based on the Duparc classification). In some of these fractures, the fragments will be seen to be holding together reasonably well, in which case minimal internal fixation with conventional means would be the preferred treatment modality.
- Fracture-dislocations (other than those involving complete loss of attachments of the humeral head in the elderly).
- Displaced infratubercular fractures in patients with osteoporotic bone stock, or with comminution of the metaphysis/diaphysis.
- Fracture patterns involving little or no displacement;
- Fracture-dislocations in the elderly;
- Fractures involving comminution of the humeral head;
- Ongoing infection.
In conclusion, the Bilboquet device permits the reduction and sound fixation of complex fractures of the proximal humerus.
Its use produces outcomes that can stand comparison with those of all the internal fixaton techniques currently recommended, and obviates the need for hemiarthroplasty.
Any avascular necrosis observed following the use of the Bilboquet tends to be well tolerated, because of the good healing behaviour of the tuberosities. While the device was designed to offer the potential of conversion to a hemiarthroplasty, there has been very little need so far to resort to this facility.
REFERENCES 1 - DOURSOUNIAN L, GRIMBERG J, CAZEAU C, TOUZARD RC. Une nouvelle méthode d'ostéosynthèse des fractures de l'extrémité supérieure de l'humérus : l'implant bilboquet. A propos de 17 cas revus à plus de 2 ans. Rev Chir Orthop Reparatrice Appar Mot. 1996;82: 743-52
2 -DOURSOUNIAN L, GRIMBERG J, CAZEAU C, JOS E, TOUZARD RC. A new internal fixation technique for fractures of the proximal humerus--the Bilboquet device: a report on 26 cases.. J Shoulder Elbow Surg. 2000;9:279-88.
3 - NEER CS II. Displaced proximal humeral fractures. Part I. Classification and evaluation. J Bone Joint Surg Am 1970;52A:1077-89.
4 - NEER CS II. Displaced fractures of the proximal humerus. Part II.Treatment of three-part and four-part fractures. J Bone Joint Surg Am 1970;52A:1090-103.
5 - WILLEMS WJ, LIM TE. : Neer arthroplasty for humeral fracture. Acta Orthop Scand 1985;56: 394-95.
6 - WRETENBERG P, EKELUND A. Acute hemiarthroplasty after proximal humerus fracture in old patients. A retrospective evaluation of 18 patients followed for 2-7 years. Acta Orthop Scand 1997;68: 121-23.
7 - ZYTO K, WALLACE A, FROSTICK SP, PRESTON JB. Outcome after hemiarthroplasty for three- and four part-fractures of the proximal humerus. J Shoulder Elbow Surg 1998;7:85-89.