Background Proximal humerus fractures (PHFs) are common in older adults. However, no classification system has achieved universal consensus. Recently, the Mayo Clinic and Fundación Jiménez Díaz (M-FJD) introduced a new classification based on pattern recognition. This classification achieved substantial interobserver and intraobserver agreement using radiographs (XR) and computed tomography (CT), with higher agreement on CT. The aim of this study was to validate this classification by assessing agreement among an independent group of evaluators. Additionally, we aimed to evaluate the disengaged pattern, an additional fracture pattern discussed—but not formally assessed—in the original M-FJD description. Methods This was an imaging-based observational study. A total of 120 patients diagnosed with PHF who underwent evaluation with XR and CT were included. The images were assessed by 3 surgeons and 1 shoulder fellow. Both XR and CT images—including axial, coronal, sagittal, and 3-dimensional reconstructions—were reviewed. All cases were classified according to the 8 patterns of the M-FJD classification, and reviewers recorded their level of diagnostic certainty for each case. Intraobserver reliability was calculated after a 6-week interval by randomizing the image sets. A third evaluation round was conducted to assess the disengaged pattern. Kappa statistics were used to determine both interobserver and intraobserver agreement. Results Interobserver agreement was substantial for XR (kappa = 0.65 [0.62–0.73]) and CT (kappa = 0.62 [0.59–0.68]). Interobserver agreement in the second assessment was again substantial for XR (kappa = 0.68 [0.63–0.73]) and CT (kappa = 0.63 [0.60–0.65]). Intraobserver agreement for each rater and overall agreement was substantial for XR (kappa = 0.76 [0.73–0.80] and CT (kappa = 0.73 [0.70–0.74]. The difference in the level of uncertainty regarding pattern recognition in XR (20.98%) vs. CT (13.34%) did not reach statistical significance (P > .05). Additionally, the inclusion of the disengaged pattern did not improve agreement between observers. Conclusion Substantial interobserver and intraobserver agreement can be expected when using the M-FJD classification for PHFs, regardless of imaging modality (XR or CT). Adding the disengaged pattern did not improve interobserver agreement in this independent validation of the M-FJD classification.
