Imaging to improve agreement for proximal humeral fracture classification in adult patient: A systematic review of quantitative studies.

Proximal humeral fracture classification has low reproducibility. Many studies have tried to increase inter- and intra-observer agreement with more sophisticated imaging. The aim of this review was to determine which imaging modality produces the best inter- and intra-observer agreement for proximal humeral fracture classification in adults and to determine if this varies with observer experience or fracture complexity. OvidMEDLINE, The Cochrane Library, EBSCO CINAHL and Elsevier Scopus were searched on July 22nd, 2018. Quantitative studies comparing at least two imaging modalities for inter- or intra-observer agreement of proximal humeral fracture classification in adults were eligible for inclusion in this systematic literature review. Two reviewers independently screened and extracted data. Study quality was appraised using a modified Downs and Black checklist. The search strategy identified 1987 studies, of which 15 met the eligibility criteria. All included studies addressed inter-observer agreement and 8 provided results for intra-observer agreement. A narrative synthesis was performed. Trends were compared between studies as clinical heterogeneity and the statistical measures used by included studies prevented meta-analysis. Inter- and intra-observer agreement was found to increase from radiographs (x-ray) to two-dimensional (2D) computed tomography (CT) to three-dimensional (3D) CT. 2D and 3D CT may improve inter-observer agreement to a greater extent in less experienced observers and in more complex fractures. Future studies should compare 2D and 3D CT with subgroups categorising surgeon experience and fracture complexity. X-ray should be used for initial assessment; however doctors should have a low threshold for ordering CT. PROSPERO number: CRD42018094307.