Comparing optimum prosthesis combinations of total stemmed, stemless and reverse shoulder arthroplasty revision rates for men and women with glenohumeral osteoarthritis.

Background: This study investigated prostheses from a large national arthroplasty registry with the lowest rates of revision, defined as optimum. We compared optimum shoulder arthroplasty revision rates for osteoarthritis (OA) to determine the most suitable/effective procedure for men and women.

Methods: There were 3 cohort groups of optimum primary shoulder arthroplasties for OA undertaken between January 1, 2008, and December 31, 2022: stemless shoulder arthroplasty with cemented polyethylene glenoids (slTSA), stemmed shoulder arthroplasty with modified central peg polyethylene glenoids (stTSA), and cementless reverse shoulder arthroplasty (rTSA). The cumulative percentage revision (CPR) was determined using Kaplan-Meier estimates of survivorship and hazard ratios (HRs) from Cox proportional hazard models adjusted for age, gender, humeral head/glenosphere size, polyethylene type, and surgeon volume. Possible interactions were examined. A subanalysis from January 1, 2017, captured additional patient demographics, American Society of Anesthesiologists score, BMI, and glenoid morphology.

Results: The CPR at 7 years was 4.0% (95% confidence interval [CI] 3.1, 5.1) for slTSA (n = 3041), 3.8% (95% CI 2.7, 5.5) for stTSA (n = 1259), and 4.1% (95% CI 3.7, 4.6) for rTSA (n = 12,341). slTSA had a higher rate of revision compared with rTSA after the first 9 months (P < .001). rTSA had a lower revision rate than stTSA from 3 months on (P = .004). After adjusting for other confounders, prosthesis type and gender were associated with revision rates (P < .001) whereas surgeon volume was not. Additionally, gender and prosthesis type strongly interacted (P = .013) and the combined model exhibited greater predictive performance when including this interaction. Women had lower rates of revision than men for both stTSA and rTSA but not slTSA. Most revisions were for infection in men, especially rTSA. After 3 months, the rate of revision for slTSA vs. rTSA for women was increased (P < .001) and revision rates for men did not significantly differ. However, in a subanalysis of procedures in males since 2017 with additional adjustments, slTSA had a lower revision rate than stTSA (P = .010).

Conclusions: The optimum shoulder arthroplasty revision rates vary for both the gender and implant type for the diagnosis of OA. A model combining optimum prostheses and gender predicted revision better than optimum implants alone. After 3 months, rTSA was associated with lower revision rates compared with slTSA in women, whereas there were no significant differences between optimum prostheses in men. However, surgeons may also consider lower revision risk of optimum slTSA at subanalysis and increased cumulative incidence of infection for rTSA requiring revision to resolve decision making for male patients.