Fig. 6

a-d - ROM analyses for the PRSA and SRSA subgroups: The Kruskal-Wallis test provided strong evidence of differences between the mean ranks of the three groups in (a) flexion (p = .020) and (b) abduction (p = .020). When results were adjusted using the Dunn-Bonferroni method, there was strong evidence of a difference between failed ORIF SRSA and PRSA (p = .038) and failed ORIF SRSA and other SRSA (p = .017) in terms of flexion. Similarly, there was strong evidence of a difference between failed ORIF SRSA and PRSA (p = .026) and failed ORIF SRSA and other SRSA (p = .034) in terms of abduction. The other results show that there was no statistically significant difference between external (c) and internal rotation (d). Flexion, abduction, and external rotation are displayed in degrees and internal rotation in points according to a sub-item of the Constant score [32]. (failed ORIF SRSA – reverse total shoulder arthroplasty following failed open reduction and internal fixation, PRSA – primary reverse total shoulder arthroplasty, ROM – range of motion, SRSA – secondary reverse total shoulder arthroplasty)