We propose an approach to open reduction and internal fixation (ORIF) of proximal humerus fractures in the elderly (older than 75 years). This combines several techniques in a “hybrid” approach, with locked plating and rotator cuff traction sutures.

All surgery was performed by the senior author (MT). A systematic approach was utilized in reducing and maintaining reduction of all fractures in this series [18]. This began with a thorough review of preoperative imaging. Preoperative CT scans were obtained in most cases to understand the fracture morphology. In addition, contralateral shoulder radiographs were obtained to aid in assessing the reduction obtained intraoperatively. Before preparing and draping the patient, the ability to obtain high-quality fluoroscopic images was ensured, including an AP in the plane of the scapula with the arm held in external rotation to profile the greater tuberosity and a Velpeau axillary with the arm held in internal rotation to profile the lesser tuberosity.

A deltopectoral approach was utilized in all cases. A large bone reduction clamp was used to manipulate the humeral shaft. Rotator cuff traction sutures were placed to allow direct control over the tuberosities and indirect control of the head through the tuberosities when not fractured. In cases where the humeral head is detached from the tuberosities, a pin may be inserted and utilized as a joystick for control of the head segment; this was seldom required in the fractures included in this particular report. The fracture was then gently reduced. In surgical neck fractures, a small malleable retractor was used as a shoehorn to aid in reduction of the shaft under the head. In valgus impacted fractures, the humeral head was gently elevated, paying careful attention to preserving the medial hinge and associated blood supply. Considerable effort was taken to ensure adequate support of the humeral head after reduction. This was performed using the shaft (valgus “impaction osteotomy” of Fenlin), or in cases of extensive metaphyseal comminution and osteopenia, bone graft or bone graft substitute was used (Fig. 1) [2]. Bone graft was utilized in 15 cases (43%) in this series. This was autograft bone in seven cases and allograft in eight cases. Next, provisional fixation was obtained using a Kirschner wire (Fig. 2). The reduction was then assessed utilizing fluoroscopy in the AP and Velpeau axillary views. Contralateral films were used to avoid varus malreduction. In all cases, the fracture was reduced, adequate support was obtained (especially medially), and provisional fixation was placed before application of the plate. Adequate medial support was defined as contact between the humeral head segment and the shaft segment at the medial calcar region, especially posteromedially. Thus, with this technique, the plate was used in the neutralization mode rather than as a reduction tool.

The plate was positioned low enough to avoid subacromial impingement. Long screws into the subchondral bone were avoided to decrease the risk of intra-articular penetration. The screws were placed by predrilling the lateral cortex and advancing the depth gauge without further predrilling to avoid placing screws close to the subchondral bone. All screws were scrutinized under fluoroscopy in varying degrees of internal and external rotation to ensure no intra-articular hardware. Rotator cuff traction sutures were utilized liberally to neutralize the varus deforming force of the supraspinatus (Fig. 3A). Because of the humeral head support obtained using the shaft, bone graft, or bone graft substitute, short screws and tension band suture fixation provided adequate stability to the humeral head fragment in all cases (Fig. 3B).

All patients were immobilized for 6 weeks after the operation. Patients were placed into a shoulder immobilizer immediately postoperatively and were instructed to stay in the immobilizer at all times until their first visit at 6 weeks. Patients were allowed to remove the immobilizer for hygiene only. Assisted ROM was initiated at 7 to 12 weeks, with therapist-driven ROM exercises and sling use the remainder of the day. After 3 months or demonstration of radiographic healing, the sling was discontinued.

A chart review was completed to determine the clinical outcome. ROM assessed at clinic appointments was gathered, and surveys were sent to all 25 living patients. The survey (Appendix 1) included illustrated self-assessments of ROM, in which patients selected diagrams showing maximum ROM, and a visual analog pain scale (1–10, with higher numbers indicating more severe pain) and allowed calculation of the American Society of Shoulder and Elbow Surgeons (ASES) score [15], the Simple Shoulder Test [12], and the QuickDash [4]. The QuickDash is a general study of upper-extremity function, with lower scores being better. The Simple Shoulder Test and ASES scores are shoulder-specific scores in which a higher score indicates better function. A second survey was sent to the 14 nonresponders within 2 months. Followup telephone calls were made to the five patients who did not respond to the second survey, and survey information was obtained via telephone.

Radiographs of the involved shoulders were assessed for union and ON. Healing was considered complete when there was evidence of both cortical continuity on all cortices and trabecular continuity in the AP radiographs of the shoulder in internal and external rotation, as well as in the axillary and scapular Y radiographs. ON was diagnosed when radiographs demonstrated subchondral bony defects with or without a crescent sign or articular collapse, as previously described [5]. All radiographs were reviewed by the leading author (JDB) and a fellowship-trained shoulder and elbow surgeon (JSS) who was blinded to the clinical outcome of the patients, and consensus was achieved. There were no cases where healing could not be assessed. Reduction was considered satisfactory when the head-shaft angle was between 40° and 55° and there was anatomic reduction of the tuberosities within 5 mm.

Satisfactory reduction was obtained in 31 of 35 fractures (89%). Four fractures were considered in excessive valgus (11%). All fractures healed. At last followup, four patients (11%) had partial ON. Two of these patients were asymptomatic and two complained of moderate pain with activity. Of these two symptomatic patients, one responded to nonoperative management, while one was offered arthroplasty but died during preoperative medical workup.

There were several limitations to the study. First, this study involved a selected series of patients without a control group. An attempt to minimize the inherent bias of this study design was made by enrolling a consecutive group of patients meeting the study criteria. Second, there was a relatively high rate of study dropout. High rates of dementia and death are inherent to the study population (patients older than 75 years). To minimize the number of patients lost to followup, all available resources were utilized to contact all patients. Two surveys were mailed and patients were called. Finally, the followup method of this study was through a survey in addition to clinical examinations. For this reason, only well-validated outcome measures were used. This was an attempt to reduce the inherent subjectivity of patient responses.

In this case series, ORIF was quite safe. Recent reports in the literature of ORIF of proximal humerus fractures have demonstrated high complication rates. Solberg et al. [16] documented a 16% reoperation rate, with 8% of these being within 24 hours of the initial operation. A large study conducted by Sudkamp et al. [17] demonstrated primary screw penetration in 14% of patients and an overall complication rate of 34%. In our study, there were no cases of reoperation. The most important steps for minimizing the possibility of this complication included ensuring adequate intraoperative fluoroscopy before preparing and draping the patient and carefully assessing each screw placed with internal and external rotation views of the joint to confirm the hardware was not intra-articular. In addition, contrary to previous reports, short screws were utilized to engage the humeral head. Previous series have used long screws penetrating into the subchondral bone of the humeral head. This may be a cause of early and delayed intra-articular screw penetration.

Fracture reduction was maintained in most of the shoulders in our study. The reduction obtained in this series was adequate in 89% of patients and led to healing in all instances. This result is in contrast to reports of high rates of varus malreduction (up to 36%), which leads to early hardware failure [16]. Intraoperative comparison with the contralateral shoulder radiographs was an important step in reducing the frequency of malreductions. In addition, adequate reduction and humeral head support were attained before placement of the plate. This is in contrast to previous reports using the plate as a reduction tool [10]. There were no cases of varus malreduction, which emphasizes the attention paid to ensuring adequate valgus alignment. In addition, previous studies have documented ON in up to 37% of patients [19]. We had four patients (11%) with ON. Two of these patients had no or minimal pain and one responded to nonoperative management. The average visual analog pain score of the operative extremity was less than 1 point higher than the contralateral (unaffected) extremity (2.5 versus 1.8), indicating reliable pain relief at last followup.

Six weeks of immobilization did not lead to disabling stiffness in this series. While prolonged immobilization may be detrimental in young patients, this patient population seems not to be prone to develop stiffness despite immobilization. Patients in this study were kept strictly immobilized. This may be one reason for the absence of nonunions in this series. The mean postoperative motion of these patients included elevation of 141° and abduction of 132°. This compares favorably with the study by Sudkamp et al. [17], which emphasized early motion (within 2 days postoperatively). Their patients averaged 132° of elevation and 122° of abduction. Both of these groups of patients have better motion than is consistently obtained in hemiarthroplasty. Recent studies on hemiarthroplasty for proximal humerus fracture have reported elevation of 100° to 105° and abduction of 92° [1, 11]. In addition to the objective motion obtained in this patient group, it should be emphasized the demands of patients older than 75 years are different from those of younger patient populations. None of our patients had a decrease in ability to perform activities of daily living secondary to their fracture compared to preoperatively.