Exactech Equinoxe anatomic versus reverse total shoulder arthroplasty for primary osteoarthritis: case controlled comparisons using the machine learning-derived Shoulder Arthroplasty Smart score.

Background: The role of reverse total shoulder arthroplasty (rTSA) for glenohumeral osteoarthritis (GHOA) with an intact rotator cuff remains unclear with prior investigations demonstrating similar patient-reported outcome measures (PROMs) to anatomic total shoulder arthroplasty (aTSA). However, legacy PROMs are subject to skewed distributions with many patients achieving the maximum possible score (ceiling scores). We evaluated a cohort of primary rTSAs performed for GHOA with an intact rotator cuff compared with a case-matched cohort of aTSAs using the Shoulder Arthroplasty Smart (SAS) score, a machine learning-derived outcome measure that eliminates the ceiling effect.

Methods: A retrospective review of an international shoulder arthroplasty database was performed between 2001 and 2020. Patients undergoing rTSA for rotator cuff intact GHOA (n = 367) were matched 1:1 with aTSA controls (n = 367) with a minimum of 2-year follow-up. Assessed variables included patient demographics, range of motion, American Shoulder Elbow Surgeons score (ASES), Constant score, Simple Shoulder Test (SST), and the SAS score.

Results: Preoperatively, the SAS (49.0 vs. 45.2; P < .001), SST (4.7 vs. 4.1; P = .002), and Constant score (42.5 ± 15.3 vs. 38.1 ± 14.2; P < .001) were greater in aTSA vs. rTSA. Similarly, the SAS score (82.3 vs. 77.6; P < .001) and SST score (10.8 vs. 10.3; P = .003) remained greater in aTSA postoperatively. In contrast, no differences in the ASES (P = .103) or Constant scores (P = .108) were found between aTSA and rTSA patients postoperatively. Improvement preoperatively to postoperatively did not differ between aTSA and rTSA patients when assessed using the SAS (P = .257), ASES (P = .888), or SST scores (P = .510). However, a higher rate of improvement in the Constant score (35.0 vs. 31.8; P = .022) was observed in rTSA compared with aTSA. Higher rates of complications (5.4% vs. 1.6%; P = .008) and revision surgery (4.1% vs. 0.5%; P = .002) were observed in aTSA compared with rTSA.

Conclusions: Patients undergoing shoulder arthroplasty for rotator cuff intact GHOA experienced higher rates of complications and revisions in aTSA compared with rTSA, but similar levels of clinical improvement between implants when evaluated using a validated outcome score without a ceiling effect. Unlike the ASES and SST scores that are limited by ceiling effects, a higher mean postoperative SAS score after aTSA was observed, but preoperative to postoperative SAS differences were similar after rTSA.