Is there a difference in thresholds for revision between shoulder arthroplasty types? A National Joint Registry Study

Abstract

Introduction Shoulder arthroplasty procedures have increased significantly, with reverse shoulder arthroplasty (RSA) becoming more common. While RSA revision rates are reported as low, these figures may not accurately reflect implant success. Factors such as older patient demographics and surgeon reluctance to perform complex revisions may contribute to lower revision rates. This perception may encourage broader RSA use in younger patients, potentially increasing long-term revision burdens. This study examines whether revision thresholds differ between shoulder arthroplasty types to determine if low revision rates reported signify true implant success. Methods All shoulder arthroplasties from the 1st April 2012 to the 31st March 2022 were requested from the National Joint Registry (NJR). Mean postoperative Oxford Shoulder Score (OSS) was calculated for RSA, total shoulder arthroplasty (TSA), and hemiarthroplasty (HA). Revision rates were analysed between implants for patients with a mean postoperative OSS of <29 (a pre-defined unsatisfactory score) and those with the lowest 25%, and lowest 10% of OSS scores. Chi-squared tests with Bonferroni correction assessed differences among implant groups. Results Among 21,918 NJR patients with postoperative OSS data, HA had the highest proportion of patients with ‘unsatisfactory’ function (38.12%), followed by RSA (26.99%) and TSA (15.35%). 4.87% of RSA patients with unsatisfactory function were revised, significantly less than TSA (10.58%) and HA (13.86%) (p < 0.001). In those with the lowest 25% of OSS scores, revision rates were 4.78% for RSA, 8.76% for TSA, and 15.02% for HA (p < 0.001). In the lowest 10%, revisions occurred in 6.53% of RSAs, 12.44% of TSAs, and 17.03% of HAs (p < 0.001). No significant difference was found between TSA and HA (p = 0.06). Conclusion RSA has lower revision rates than HA and TSA; however, this may not reflect superior implant performance. Patients with poor RSA function are less likely to undergo revision, suggesting higher revision thresholds. As RSA use expands, its assumed low revision rate must be reassessed to prevent long-term burdens of poorly functioning implants. Further research is needed to determine whether surgical selection bias influences revision rates and to establish additional benchmarks or surrogates in joint registries for a more comprehensive assessment of implant performance.