Patient-evaluated functional outcome is systematically different from physician-evaluated outcome after proximal humerus fracture fixation

Aims The study aimed to investigate the agreement between Constant-Murley score (CMS) and the shoulder subjective value (SSV); and the association of poor self-evaluated outcome with shoulder pain, range of motion (ROM) limitation, and shoulder strength after fixation of proximal humerus fractures. Methods The study included 74 shoulders in 72 patients (mean age 56 years (SD 14), 40 male (56%)) who underwent locking plate fixation of displaced proximal humerus fractures between January 2015 and January 2024. We evaluated CMS, SSV, pain scores (visual analogue scale), and ROM measurements at the most recent (minimum one-year) follow-up. Agreement statistics between CMS and SSV were used descriptively, rather than to test true substitutability, because CMS and SSV are nonequivalent constructs. Results The Bland-Altman analysis showed differences between CMS and SSV, as the mean difference between CMS and SSV (-14.21 (95% CI -17.25 to -11.18)) did not include the zero-difference line. The Pitman’s test suggested that the variances were significantly different (p = 0.029). Firth’s penalized method of logistic regression revealed that only the presence of shoulder pain was associated with an increased likelihood of a poor functional outcome (SSV ≤ 69.4) (coefficient 3.5 (95% CI 1.53 to 6.47); p < 0.001). The SSV threshold (69.4) was derived from the CMS ≤ 55 threshold on the same dataset and therefore may have nonindependence. However, a sensitivity analysis using a priori SSV threshold (75) of patient-acceptable symptom state (PASS) also showed that presence of pain was associated with increased likelihood of a lower SSV. Limitations in ROM (elevation, external rotation, and internal rotation) and lower shoulder strength were not independently associated with a poor SSV. In shoulders that did not develop avascular necrosis (n = 66), the bivariate exploratory analysis between presence of pain and predefined injury and radiological factors showed that malreduction (χ 2 (1, n = 66) = 6.8; p = 0.009) and absent postoperative medial hinge (χ 2 (1, n = 66) = 4.6; p = 0.032) were significantly associated with more frequent presence of shoulder pain. Conclusion Physician-evaluated outcomes (as measured by the CMS) and patient self-evaluated outcomes (as measured by the SSV) are likely to differ systematically after internal fixation of proximal humerus fractures. Additionally, shoulder pain is independently associated with a poor self-evaluated functional outcome. Cite this article: Bone Joint J 2026;108-B(9):1162–1168.