Reverse shoulder arthroplasty (RSA) has become a widely used procedure since its introduction in the 1980s, and is currently used to treat a wider range of conditions than its original indication. The original Grammont-style RSA revolutionized shoulder arthroplasty but had several limitations, including scapular notching and reduced rotational motion. This review discusses the evolution of RSA design, particularly the development of a lateralized center of rotation constructs, which aims to improve all the disadvantages associated with the Grammont-style design and more closely reproduce the native anatomy in order to improve patient outcomes in an expanded context of pathologies.

The introduction of different humeral implants for the treatment of osteoarthritis was intended to improve the ability to anatomically reconstruct the proximal humerus. New technology should ideally lead to improvements in optimal reconstruction. The primary aim of this study was to compare three humeral implant types (long stem, short stem, and stemless) to replicate the proximal humeral anatomy and assess the impact of transitions across each stem type, including initial learning curves. This was a retrospective review of 298 patients (175 males and 123 females) who underwent anatomic total shoulder arthroplasty utilizing three following stem types: 145 long stem, 102 short stem, and 51 stemless implants. An AP radiograph which met a criterion of optimal/orthogonal view of the humeral head and stem was selected for every included patient. A best fit circle, an articular surface circle and postoperative measure of neck-shaft angle (NSA) was measured by two independent observers. The distance between the centers of two circles (center of rotation [COR] shift) and NSA were then compared for each stem type. In order to assess transitions in technology, two sets of patients in the stemmed groups were selected: initial year (79 long stem, 62 short stem) and final year (66 long, 40 short) of utilization. A cumulative sum control chart analysis was used to assess the learning curves of each of the stem types by the initial year of utilization according to the radiographic measurement of COR shift. The stemless implant showed best reconstruction with a mean COR shift of 3.0 ± 1.6 mm and NSA of 137 ± 6° (P < .001). The transition from long stem to short stem utilization showed significantly worse COR shift and NSA, 3.3 ± 2.1 mm to 4.1 ± 1.9 mm and 138.5 ± 4.4° to 141.0 ± 4.4° (max P = .032), respectively. The impact of transitions to newer technology demonstrated an increased operative time (100 to 128 and 135 min). The cumulative sum control char learning curves demonstrated a completion of the learning phases at 12 cases for both the long and short stem designs, but there was no identifiable learning phase for the stemless implant, suggesting an immediate entry to the consolidation phase. When assessing reconstruction of the premorbid shoulder joint, the stemless implant had the greatest efficacy and shortest learning curve. Stemless anatomic humeral reconstruction is more replicable in recreating the anatomic shape of the humerus and is quicker to master compared to standard or short stemmed implants but did have a greater initial operative time.