Factors associated with failure of locking plate fixation in proximal humerus fractures

Background: Locking plate fixation remains the mainstay of surgical fixation of unstable proximal humerus fractures, however rates of failure remain high. The aim of this study was to identify risk factors that could be used to predict the likelihood of fixation failure. Method: Patients with proximal humerus fractures managed with locking plate fixation between 2010 and 2019 at a Level 1 trauma centre were included. Radiographs were evaluated for parameters that could be used to predict failure of fixation. Pre-operative factors included were the Neer classification, cephalomedullary angle, medial calcar length, disruption of the medial hinge, and anatomical neck fracture. Post-operative factors included the cephalomedullary angle, medial calcar reduction gap, presence of anatomical tuberosity reduction, presence of medial calcar screws, screw distance to articular surface, and number of screws present in the humeral head. Results: There were 189 patients included; 54 % male, mean age 49.9 (intact fixation) group and 56.1 (failure). The rate of fixation failure was 22 %. Factors associated with increased risk of failure following multivariable analysis included increasing age (OR 1.04 per year, CI 1.01–1.07), varus pre-operative cephalomedullary angle (OR 2.84, CI 1.03–7.83), and non-anatomical calcar reduction (OR 2.31, CI 1.05–5.08). The presence of calcar screws was associated with decreased risk of fixation failure (OR 0.30, CI 0.10–0.90). This analysis was used to create a predictive model including the Neer classification, age, pre-operative cephalomedullary angle, post-operative cephalomedullary angle, anatomic reduction of the medial calcar, and presence of medial calcar screws. Conclusion: Rates of locking plate fixation failure in proximal humerus fractures remain high. This study has identified key pre-operative and intra/post-operative factors that can be used to predict the risk of failure. Further work is required to validate this model.