Text

Objectives and questions: Despite favorable average outcomes, postoperative recovery after reverse shoulder arthroplasty (RSA) remains heterogeneous, and mean score–based reporting may obscure clinically relevant outcome. The purpose of this study was to identify distinct postoperative recovery phenotypes following RSA for cuff tear arthropathy (CTA) and to evaluate associations between baseline patient- and surgery-related factors and phenotype membership.

Material and methods: This retrospective cohort study analyzed a prospectively collected shoulder arthroplasty registry. Patients undergoing primary RSA for CTA with complete 24-month follow-up were included. Unsupervised k-means clustering of 24-month Subjective Shoulder Value (SSV), active shoulder abduction, and Numeric Rating Scale (NRS) pain scores identified postoperative recovery phenotypes. Multinomial logistic regression evaluated associations between baseline demographic (age, ASA class, sex, BMI, smoking, alcohol abuse), clinical, and surgical factors (implant design, rotator cuff tear severity) and cluster membership, using the High-Recovery phenotype as reference.

Results: 518 patients were included. Using unsupervised k-means clustering three recovery phenotypes were identified. The High-Recovery phenotype comprised 287 patients (55%) and was characterized by high functional outcomes (mean SSV 88.3), greater active abduction (143°), and minimal pain (mean NRS 0.66). The Intermediate- Recovery phenotype included 153 patients (30%) with moderate function (mean SSV 81.3), lower abduction (102°), and slightly higher pain levels (mean NRS 0.99). The Low-Recovery phenotype consisted of 78 patients (15%) and demonstrated substantially worse outcomes, with low SSV (51.5), limited abduction (107°), and high pain burden (mean NRS 4.94). Multinomial logistic regression was performed to evaluate associations between baseline-, surgery- and recovery-phenotype membership, using the High-Recovery phenotype as the reference. ASA class was independently associated with Intermediate-Recovery outcomes (OR 0.39, 95% CI 0.21–0.70; p = 0.001), but not with Low-Recovery outcomes (p = 0.30). Alcohol use was significantly associated with both Intermediate-Recovery (OR 3.08, 95% CI 1.51–6.24; p = 0.002) and Low-Recovery outcomes (OR 2.75, 95% CI 1.25–6.04; p = 0.012).

Discussion and conclusions: Three distinct and clinically meaningful postoperative recovery phenotypes were identified. Recovery phenotype membership was primarily associated with patient- related factors, particularly comorbidity burden and alcohol use, whereas age, body mass index, rotator cuff tear severity, and implant design demonstrated no significant associations. Other soft-tissue and scapulothoracic related factors may be relevant and have to be included in the future for phenotype-based outcome assessment and patient counseling, expectation management, and risk stratification may facilitate more individualized decision-making in shoulder arthroplasty.