Shoulder hemiarthroplasty (HA) began its modern development in 1951 when Neer designed his original monoblock prosthesis to address unsatisfactory results after operative fixation of complex proximal humerus fractures. His first published clinical series of 13 cases included one case for treatment of degenerative arthritis.¹ Several years after its introduction, the system was expanded to include multiple stem diameters, though there was only one humeral head diameter of 44 mm. Neer used this system for over 20 years and reported “When there was a good rotator cuff, a good deltoid muscle, and a good rehabilitation regimen, an excellent result could be obtained.”²

Over the next 3 decades, humeral prosthesis design evolved to add progressive modularity and adaptability in an effort to better approximate proximal humeral anatomy.³,⁴ Now in their fourth generation, systems have included the ability to independently adjust the anterior/posterior and medial/lateral head offset as well as platform stems that offer easy convertibility between anatomic and reverse arthroplasty. Resurfacing implants, developed originally in Scandinavia in the early 1980s and popularized by Copeland, have been used as an alternative to traditional stemmed implants for HA. They offer the advantages of bone preservation and the ability to address altered anatomy between the humeral head and shaft such as occurs in fracture malunions. Stemless implants have also been more recently introduced, also avoiding potential problems associated with stemmed implants including stress shielding, periprosthetic fracture, and the difficulty associated with stem removal at revision surgery. All of these design options can be applied to shoulder HA, and the advantages and disadvantages will be discussed.

The incidence of all shoulder arthroplasty has increased substantially over the past 20 years. Data from the Australian and New Zealand Joint Replacement Registries indicate that the number of shoulder replacement procedures has increased over 150% in the past 2 decades.⁵,⁶ This growth is driven by several factors which include overall population growth, demographic changes, expanding indications, and improvements in implant design. The increased demand for treatment of arthritis in younger patients has also been documented. Padegimas et al have reported an 8% annual increase in shoulder arthroplasty in patients younger than 55 years and predicted a 333% increase in demand between 2010 and 2030.⁷ Even Neer’s original series of arthroplasty for treatment of glenohumeral arthritis, published in 1974, had an average age of 55 years at the time of surgery.⁸

During this same period, shoulder HA has seen a progressive decline in utilization. The Australian Registry reported that HA as a percentage of all shoulder replacement declined from 32% to 4% between 2008 and 2018 and that the percentage of HA performed for osteoarthritis has decreased from 60% to 25% over the same period.⁵ FIGURE 19.1 shows a histogram of arthroplasty type between 1999 and 2018 from the New Zealand Joint Registry. This shows a dramatic decline in HA and resurfacing arthroplasty from 47% to 7% of cases.⁶ The Kaiser Permanente Registry has demonstrated that HA accounted for only 34% of all shoulder arthroplasty with only 38% (13% of the total) of these performed for arthritis.⁹

HA was traditionally used for a variety of diagnoses, including fractures, fracture sequelae, osteoarthritis, rheumatoid arthritis, osteonecrosis, and cuff tear arthropathy. Some portion of the decline in HA utilization is a result of the increasing use of reverse shoulder arthroplasty for several of the diagnoses for which HA is becoming more of a historical option or a salvage procedure. The use of HA for primary glenohumeral osteoarthritis has also declined based on literature suggesting superior pain relief and functional outcomes of total shoulder arthroplasty (TSA) as will be discussed further. Nevertheless, there are a subset of patients for whom HA may still be an appropriate treatment option in the management of degenerative joint disease. This chapter will focus on the role of HA for the treatment of glenohumeral osteoarthritis in this subset of patients. Biological resurfacing with HA will not be discussed since this technique has largely fallen out of favor.¹⁰,¹¹,¹²

In order to achieve a favorable outcome with HA, surgeons must adhere to principles and indications that recognize those clinical scenarios for which this option may be a comparable choice to TSA. The most common current indications for HA include select cases of primary osteoarthritis with an intact rotator cuff, atraumatic osteonecrosis, and select cases of secondary shoulder arthritis such as postcapsulorraphy arthropathy. In the cuff-intact shoulder, HA tends to be favored in younger patients or those with functional demands which may jeopardize the long-term survival of TSA.

Thus, a principle rationale for HA is the anticipation of glenoid implant failure and its consequences for salvage options in patients with a remaining life expectancy which is likely to exceed that of their implant. Several authors have noted a high frequency of radiolucent lines after TSA with a consequent negative impact on functional outcomes. Pfahler et al reported 68% glenoid radiolucent lines at an average 4-year follow-up,¹³ while Galluser and colleagues noted 52% radiolucent lines with 11% recurrent posterior subluxation in patients with preoperative retroversion and decentering of the humeral head.¹⁴ Roberson et al reported a 54% lucency rate in patients under age 65 with a 17.4% revision rate most commonly due to glenoid component loosening.¹⁵ Similarly, Edwards reported 56% of glenoid implants with radiolucent lines at midterm follow-up of TSA.¹⁶ Schoch and coworkers found that increasing grade of radiolucency was significantly associated with worse postoperative function.¹⁷ Walch et al have also noted worse outcomes when TSA is used to treat patients with B2 glenoids.¹⁸,¹⁹

These clinical findings are supported by biomechanical and finite element data. Farron et al showed that glenoid retroversion greater than 10° substantially increased stresses within the cement mantle and glenoid bone and resulted in increased micromotion at the bone-cement interface.²⁰ Nyffeler has also shown that retroversion is associated with posterior displacement of the humeral head and asymmetrical posterior loading of the glenoid implant which may result in loosening.²¹ While the clinical implications of radiolucent lines around the glenoid implant remain debatable, the known association between radiolucencies and implant loosening in lower extremity arthroplasty would suggest that these glenoids are “at risk” of eventual failure. This information must be carefully considered when assessing patients’ options for surgical management of glenohumeral osteoarthritis.

The most common reason for failure of HA is painful glenoid erosion. This has been documented in multiple clinical series, careful analysis of which may indicate an
association with certain preoperative factors and other issues which can be addressed through surgical technique and perioperative management. Levine et al have noted a higher revision rate for HA in patients with eccentric versus concentric wear patters.²²,²³ Norris and Iannotti also reported worse results with HA in patients with preoperative glenoid erosion and posterior humeral subluxation.²⁴ Hackett et al, reviewing a series of failed HA, also found that glenoid erosion was more common in decentered hemiarthroplasties and those positioned in valgus.²⁵ Herschel et al similarly found valgus implant position as a risk factor for postoperative glenoid erosion.²⁶

Furthermore, while glenoid erosion after HA is linked to pain and poor outcomes, it is unclear from the literature if poor outcomes contribute to erosion or if erosion leads to poor outcomes. Neer observed, “Erosion of the glenoid does occur if the shoulder is stiff, causing the articular surfaces to bear constantly against each other in one area.”² Sperling et al, in a long-term comparative analysis of HA and TSA, noted glenoid wear in 72% and further noted that stiffness was associated with unsatisfactory outcomes.²⁷ Getz and colleagues also reported worse functional results after HA in patients with residual stiffness, particularly loss of external rotation.²⁸ Because this is associated with obligate posterior humeral decentering in the native arthritic shoulder, postoperative external rotation stiffness may be associated with recurrent asymmetric posterior loading of the unresurfaced glenoid resulting in wear and pain.

Other authors have noted worse postoperative outcomes for HA in patients who have undergone prior surgical procedures.²⁵,²⁹,³⁰ Sperling noted a higher clinical failure and revision rate for patients with a preoperative diagnosis of posttraumatic arthritis or sequelae or trauma.³¹ Hasan, characterizing risk factors for failed arthroplasty, noted fracture sequelae, prior surgery, implant malposition, and stiffness as causes of unsatisfactory outcomes.³² The collective implication from analysis of these clinical series suggests that the characteristics of HA failure are often of the type that can be minimized by proper patient selection, optimal surgical technique, and perioperative management. Each of these factors will be considered.

As Matsen has suggested, HA with or without nonprosthetic glenoid arthroplasty is “not for every surgeon, every patient, or every problem.”³³ Achieving consistent outcomes requires a full understanding of the interplay between patient-specific variables and the underlying shoulder pathology. As most surgeons have come to understand, any well-done operation can be foiled by issues such as unmet expectations, poor compliance, psychosocial problems, and comorbid conditions. Shoulder HA is no exception to this rule.

As with any arthroplasty procedure, pertinent aspects of the history as they relate to the patient’s development and experience of shoulder arthritis should be obtained. It is critical to understand the patient’s desired activities after surgery and the degree to which the arthritis is currently disabling their participation. Furthermore, careful assessment of patient expectations for pain relief and function must be determined. Patients who are expecting complete pain relief may not be suitable candidates for this procedure. Alternatively, patients who are willing to accept some degree of residual discomfort in return for eliminating the risk of glenoid implant failure may be quite satisfied with their overall improvement despite some persistent symptoms. Thus, considering margin for improvement is essential as patients who have more preoperative functional disability are more likely to be satisfied than those who still maintain acceptable and manageable levels of discomfort and dysfunction prior to surgery. Matsen has shown that patients with lower baseline SST scores have better outcomes than those with higher baseline scores, which may indicate that patients with continued well-compensated comfort and function are less likely to be satisfied with the results of surgery.³⁴ Mahony and colleagues have also reported an association between higher preoperative ASES score and less satisfactory outcomes.³⁶

Any history of prior surgery needs to be carefully evaluated in terms of its impact on anatomy, function, and surgical approach. The location of surgical incisions and their impact on the prospective arthroplasty should be noted. The presence of any existing hardware should be determined if special instruments are required to remove an implant. A history of prior injections should also be noted as both this and previous surgery may elevate a patient’s risk for infection and Cutibacterium acnes colonization. Because conventional laboratory workup for infection is often normal in the case of C. acnes, an index of suspicion along with plans for intraoperative assessment is essential.

Physical examination should focus on preoperative range of motion and cuff integrity. Preoperative external rotation is perhaps the most important measure because of its effect on obligate posterior humeral decentering and the importance of being able to restore this range at the time of surgery. Examination of passive range of motion with the patient supine on the table eliminates some of the compensation for stiffness that can occur through the scapulothoracic articulation and may provide a more accurate delineation of isolated glenohumeral range. Visible and palpable cuff muscle atrophy without substitution patterns on manual resistance testing may indicate disuse weakness, which will require significant rehabilitation to overcome. Tenderness over the acromioclavicular (AC) joint should also be assessed as some patients presenting with glenohumeral arthrosis are also at risk for symptomatic AC arthrosis, particularly weightlifters.