Primary osteoarthritis results in bony proliferation and loss of articular cartilage of both the humeral head and the glenoid. Synovial thickening and inflammatory episodes result in soft-tissue fibrosis. Bony proliferation and osteocartilaginous bodies increase joint volume, resulting in loss of motion. Commonly, the anterior capsuloligamentous complex (CLC) becomes so taut that the subscapularis eventually loses extensibility. Progressive anterior CLC/subscapularis constriction results in posterior humeral head migration, increased glenoid wear, posterior capsule attenuation, and posterior subluxation. Global loss of motion results in contracture of the rotator cuff muscle but also of the latissimus dorsi, teres major, and pectoralis major. Capsuloligamentous and musculotendinous contracture can impair postoperative outcome and expectations (4). Iannotti (4) found that patients treated with hemiarthroplasty with preoperative external rotation of equal to or less than 10 degrees had less active postoperative external rotation (25 degrees) compared with those who had greater than 10 degrees (47 degrees). They did not find the degree of preoperative internal rotation contracture affected postoperative ROM if the patient had a total shoulder arthroplasty (43 degrees versus 50 degrees). Loss of preoperative elevation did not influence postoperative outcome in either arthroplasty group.

Fortunately, most patients with advanced primary osteoarthritis do not have rotator cuff tears. Thus, dynamic stabilization is typically good. Norris and Iannotti (5) found only 9% of shoulders with primary osteoarthritis had rotator cuff tears but all were confined to the supraspinatus tendon and all were repairable. The presence of a repairable rotator cuff tear did not affect outcome, although better active external rotation was found in patients who were treated with a total shoulder arthroplasty versus a hemiarthroplasty (4). Better active ROM is expected when arthroplasty is performed on patients with primary osteoarthritis and an intact rotator cuff compared with patients with rheumatoid arthritis, posttraumatic arthritis, and rotator cuff arthropathy (5, 6, 7, 8).

The influence of posterior glenoid erosion and posterior subluxation on postarthroplasty outcome was also evaluated by Iannotti and Norris (4). Patients treated with a hemiarthroplasty having moderate to severe preoperative glenoid erosion had significantly less active external rotation compared with those treated with a total shoulder arthroplasty—16 degrees versus 32 degrees, respectively. There was also a difference between active forward elevation in patients having moderate to severe glenoid erosion treated with hemiarthroplasty (117 degrees) compared with total shoulder arthroplasty (139 degrees). Although ROM was affected, no difference was seen in pain relief or the American Shoulder and Elbow Surgeons (ASES) score. The amount of glenoid erosion did not influence the outcome of total shoulder arthroplasty as a whole. The presence of preoperative posterior subluxation resulted in less ROM, increased pain, and lower ASES scores in patients treated with arthroplasty (4).

Rheumatoid arthritis not only attacks the articular cartilage but also the glenohumeral soft tissues, including the rotator cuff tendons, long head of the biceps tendon, and subacromial bursa. Approximately 20% to 50% of all patients with rheumatoid arthritis acquire a full-thickness rotator cuff tear (9). Approximately 30% of patients with rheumatoid arthritis having a total shoulder arthroplasty will have a full-thickness rotator cuff tear (3,10), and many have significant tendon thinning. In late-stage rheumatoid arthritis, approximately 60% of patients have a rupture of the long head of the biceps tendon. This compromise to the dynamic glenohumeral stabilizers will affect postoperative outcome in patients with rheumatoid arthritis (3,10,11,12). Loss of rotator cuff function or an irreparable rotator cuff tear are relative contraindications for a total shoulder arthroplasty. Therefore, a hemiarthroplasty is performed and limited goals are expected in many patients with rheumatoid arthritis. Although limited functional motion is expected in these patients, significant pain relief is achieved.

Patients having arthroplasty for three- and four-part fractures will be discussed separately. The indication for arthroplasty in the patient with acute fracture is quite different from other patients treated with arthroplasty. These patients have spontaneous obliteration of the humeral head compared with slow destructive changes seen in various forms of arthritis. A hemiarthroplasty usually is performed for acute fracture because the glenoid usually is spared. The technical demand on the surgeon is greater because proper prosthetic placement and accurate and secure tuberosity fixation must be achieved. Security of the tuberosity fixation will determine the rate of postoperative rehabilitation progression. Evidence of tuberosity motion at the time of surgery or in early postoperative x-rays will limit early mobilization of the shoulder. It is critical for the surgeon to relay concerns of stable fixation to the therapist to prevent fracture line stress.

Tuberosity and prosthetic component malposition are related to poor functional outcomes (13,14). Boileau and colleagues (13) reported that significant tuberosity and humeral component malposition negatively influence outcome
in patients treated with hemiarthroplasty following three- and four-part fractures. They found early radiographic evidence of tuberosity malposition in 27% of patients and final tuberosity malposition in 33 of the 66 cases. Bigliani and colleagues (15) had similar findings in patients with failed hemiarthroplasty performed for proximal humeral head fractures. They reported tuberosity displacement or malunion and component malposition occurring in 59% and 42% of the failures, respectively. A “proud” humeral component results in relative lengthening of the humerus. This creates excessive tension on the rotator cuff, resulting in a possible nonunion between the greater tuberosity and the humeral diaphysis or rotator cuff stretch or tearing (13). Greater than 40 degrees of humeral component retroversion results in greater tuberosity posterior migration as a result of excessive tension on the tuberosity when the arm is internally rotated (placed in sling). Tuberosity or humeral component malposition ultimately results in rotator cuff deficiency, internal impingement, and pain (13).

Nerve lesions are commonly seen following proximal humeral head fractures. Visser and associates (16) reported that 67% of patients with proximal humeral head fractures have nerve injuries that diminished their functional outcome. The axillary nerve is most often affected, but combinations of nerve lesion were found. A nerve lesion should be searched for preoperatively, but difficulty in examining the painful shoulder may make this difficult. Any patient undergoing an arthroplasty for an acute fracture should be thoroughly examined for nerve lesions prearthroplasty and postarthroplasty.

It is important for the rehabilitation specialist to understand that reported postoperative ROM will be less in patients treated with hemiarthroplasty for three- and four-part fractures than for osteoarthritis and avascular necrosis. Boileau and colleagues (13) reported that elevation was 101 degrees, external rotation was 18 degrees, and internal rotation was to L3. Prakash and associates (17) reported that patients treated with hemiarthroplasty for three- and four-part fractures had elevation, external rotation, and internal rotation of 93 degrees, 24 degrees, and L1, respectively. They found no significant difference in outcome if operated within 30 days of injury or a mean of 13 months postinjury. Green and colleagues found that 50% of their patients treated for acute fractures could perform above shoulder level activities (18). Zyto and associates (19) reported on disappointing active elevation, averaging 70 degrees, in patients treated with a hemiarthroplasty within 4 weeks of a three- or four-part humeral head fracture. Attempts to perform aggressive ROM exercises following hemiarthroplasty resulting from fracture can result in iatrogenic injury, pain, and patient/therapist frustration. Frankle and colleagues (20) felt that too much passive external rotation motion, approximately 50 degrees, performed early postsurgery, could disrupt fracture healing. The rehabilitation specialist must be aware of the reported outcomes and healing concerns so that inappropriate therapy is avoided.

Most often patients with posttraumatic fracture arthritis have had previous humeral head fractures with or without surgical fixation. Dramatic soft-tissue scarring exists in this population in addition to distortion of normal bony anatomy as a result of heterotopic bone formation and malunion. Chronic nerve injuries may also exist and will influence outcome (16). Osteoarthritis occurs in 25% and 64% of patients with 3-part and 4-part fractures, respectively (19). Tuberosity malposition subsequent to injury or previous surgery can result in chronic rotator cuff deficiency. Postoperative ROM will be negatively influenced by presurgery contracture and poor dynamic stabilization by the rotator cuff. Torchia and associates (21) reported on their long-term results (mean follow-up 12.2 years) of patients treated with total shoulder arthroplasty. Of the nine shoulders having posttraumatic arthritis in which grading was done, approximately 45% undergoing a full exercise program had unsatisfactory results and 50% of those with limited goals were unsuccessful. The posttraumatic group achieved less range of active elevation, external rotation, and functional internal rotation.

The results of prosthetic replacement for three- and four-part fractures malunions do not approach those of prosthetic treatment in similar acute fractures (22,23). Better pain relief is achieved in patients with four-part fractures treated early with arthroplasty compared with delayed surgical intervention (23). Antuna and associates (22) reported on 50 shoulders with proximal humeral malunions who were treated with hemiarthroplasty and total shoulder arthroplasty. They found that patients requiring tuberosity osteotomy were at greater risk of having a poor result. Those who had a tuberosity osteotomy had active elevation of 89 degrees versus 119 degrees in those who did not undergo osteotomy. Hasan and colleagues (14) found that approximately 50% of the failed hemiarthroplasties were in patients treated for acute or chronic proximal humeral head fractures.

A related group of post-traumatic arthritis occurs in patients with primary glenohumeral instability. Dislocations and/or surgical intervention for instability can result in soft tissue imbalance and eventual articular cartilage destruction. Hovelius reported a 20% incidence of “dislocation arthropathy” in a 10-year follow-up of patients less than 40 years old with anterior instability (24). However, Hovelius found that the frequency of dislocation was not related to the development of arthritis. In fact, arthritis could develop after a single event. More commonly, arthritis as a result of instability is a result of
surgical intervention or “capsulorraphy arthropathy” (8,24). Neer felt that arthritis developed as a result of recurrent dislocation when the “standard” surgical procedure intended for unidirectional instability was used in the patient with multidirectional instability. He observed that patients would become unstable in the opposite direction of the surgical procedure (8). Even moderate to severe arthritis can develop with any stabilization procedure (25). If the tissue is overtightened or significant scar develops, the humeral head is displaced in the opposite direction. Anterior stabilizing procedures like a Putti-Platt or Magnuson-Stack may result in a severe internal rotation contracture, posterior humeral head subluxation, and posterior glenoid erosion. Laboratory studies support the clinical picture that an internal rotation contracture results in altered humeral head translation and arthritis (26,27).

Patients treated for capsulorraphy arthritis usually are younger (8,28) . Green and colleagues reported on 19 patients who had arthroplasty performed secondary to capsulorraphy arthritis and found a thickened anterior CLC and subscapularis tendon. They noted this tissue was denser than similar tissue found in patients with primary osteoarthritis and one-half of patients required a subscapularis lengthening procedure (28). In their series, 20% required posterior bone grafting or eccentric reaming to remedy the excessive posterior glenoid erosion. They reported an average of a 21-degree increase of elevation, thereby achieving an average of 120 degrees; external rotation averaged 41 degrees with internal rotation to T12. Patients having a subscapularis lengthening procedure had greater external rotation, 46 degrees, compared with those who did not, 27 degrees (28).

Unlike most arthritis that initially involves the articular cartilage, osteonecrosis begins in the subchondral cancellous bone, which fails and alters the articular cartilage. Osteonecrosis is classified as traumatic, related to fractures of the proximal humerus, or atraumatic, related to disease processes. When related to fractures, the wider the displacement of the articular segment from the shaft, the greater incidence of osteonecrosis (29). Multiple processes result in atraumatic osteonecrosis including systemic use of steroids, alcohol ingestion, and Cushing’s disease. From a rehabilitation perspective, osteonecrosis is characterized by good rotator cuff and deltoid function and relative preservation of bone. Therefore, prognosis and rehabilitation potential are excellent

Postoperatively, this group of patients is often expected to have limited goals rehabilitation because of irreparable rotator cuff insufficiency or extensive bone loss. Zuckerman and colleagues found that mean elevation was only to 86 degrees after hemiarthroplasty, but pain relief was significant. They also found increased strength in this group following arthroplasty but concluded this was a result of pain reduction (30). Williams and Rockwood reported on 20 patients who underwent hemiarthroplasty for glenohumeral arthritis combined with an irreparable rotator cuff (31). Using the Neer limited goals criteria they obtained 86% satisfactory results. They found active elevation improved from 70 degrees preoperatively to an average of 120 postoperatively. Arntz and colleagues reported average active elevation of 112 degrees in 21 shoulders treated with hemiarthroplasty for rotator cuff arthropathy (32). It is interesting that in four of the six shoulders that did not have a previous rotator cuff repair performed, all had greater than 135 of active elevation. None of the shoulders with previous rotator cuff repair had greater than 126 degrees. In fact, only 4 of 11 had greater than 100 degrees.

When considering all patients undergoing shoulder arthroplasty, two groups of patients emerge: standard goals and limited goals. The standard goals group has competent rotator cuff and deltoid muscles that allow full function and near normal motion. The limited goals patient has rotator cuff or deltoid deficiency or unreconstructable bone. Pain relief is always the goal in both groups, but in the limited goals group stability is emphasized over motion. This group achieves less satisfactory function than the standard goals patients. See Table 17-1.

Stiffness is difficult to define because some limitation of motion (compared with normal) following arthroplasty is not uncommon. Hasan and colleagues reported on 139 patients with failed arthroplasty surgery (14). They found that stiffness was the major complication in 69% and 78% of total shoulder arthroplasties and hemiarthroplasties, respectively. Neer also noted the relationship of failed arthroplasty and the prevalence of stiffness as a result of scarring and adhesions (33). Assuming that the arthroplasty was performed anatomically, immediate postoperative passive ROM is good, and the soft-tissues are adequate, early rehabilitation may influence the final result considerably.

Cofield reviewed 22 reported series (1,183 operated shoulders) of total shoulder arthroplasties and found a 0.6% rate of nerve injuries (11). In their 20-patient series review involving 498 hemiarthroplasties, there was a 1.6% incidence of nerve injury (11). The higher frequency related to hemiarthroplasty is most likely attributed to nerve injury during the trauma of fracture that was then identified postsurgically. It is difficult to evaluate a postfracture patient for true nerve injury because of pain and concern about maintaining fixation. In Cofield’s own series of postsurgical complications in individuals following total shoulder arthroplasty, he found that brachial plexopathy occurred in 3% of his patients and was an early postoperative finding (11,34).

Wirth and Rockwood identified 14 nerve injuries following total shoulder replacement (34). They felt that most of these injuries represented neurapraxia and would recover over time. They found that six of the nerve injuries involved the axillary nerve, three involved the ulnar nerve, two involved the musculocutaneous nerve, one involved the median nerve, and two were general brachial plexopathy. Resolution was complete in seven, incomplete in two, and did not occur in one. The extent of recovery was not defined in four. Lynch reported on a 4.3% of neurologic deficit following total shoulder arthroplasty in 117 arthroplasties. Thirteen of these involved the brachial plexus, and all were felt to be traction injuries (34).

In 2001 Visser and colleagues reported a 67% incidence of nerve injury following three-part and four-part fractures (16). They found denervation of the axillary nerve most common, but a combination of nerve injuries frequently occurred. One might assume that many of these patients went on to have arthroplasty, and the nerve injury affected outcome. However, the reported incidence of nerve injury following arthroplasty for acute fractures is much lower than what Visser’s study might suggest.