Detection of Intra-Articular Screw Penetration of Proximal Humerus Fractures: Is Postoperative Computed Tomography the Necessary Imaging Modality?

Abstract

Screw penetration is the common complication of proximal humerus fractures treated with locking plates. This study compared postoperative plain radiography to computed tomography (CT) for their abilities in determining screw penetration, and was to evaluate whether advanced imaging modalities (two-dimensional [2D] CT; three-dimensional [3D] CT) could increase surgeons' level of confidence regarding their diagnoses. Two observers reviewed radiological images of 134 patients who sustained proximal humerus fractures treated with locking plates. The observers were asked to answer two questions: (1) Is there screw penetrating into glenohumeral joint for this patient (Yes/No)? and (2) On a scale from 0 to 10, how confident are you about this diagnosis: (0-10) (0 = not at all confident; 10 = very confident)? Three evaluations were performed: (1) plain radiography alone, (2) radiography and 2D CT 4 weeks later, and (3) radiography in combination with 2D and 3D CT after that. This process was then repeated for intraobserver analysis. CT obtained almost perfect interobserver and intraobserver agreement (0.818-0.961), which was higher than radiography (0.377-0.655). For incidence of screw penetration, the significant difference was found between radiographs and CT images (P< .0125), but not between 2D and 3D CT images (P > .05). For confidence of diagnosis, the differences between imaging modalities were significant (all P< .001). We suggest that postoperative CT scans (especially 3D CT images) should be used to evaluate the intra-articular screw penetration of proximal humerus fractures, especially when surgeons have not enough confidence in determining screw penetration using radiography alone.