Observer variation for radiography, computed tomography, and magnetic resonance imaging of occult hip fractures

Background: Conventional radiography is insufficient for diagnosis in a small but not unimportant number of hip fractures, and secondary imaging with computed tomography (CT) or magnetic resonance imaging (MRI) is warranted. There are no convincing observer variation studies performed for conventional radiography or CT in occult fractures, and no large materials for MRI. Purpose: To assess observer variation in radiography, CT and MRI of suspected occult, non-displaced hip fractures, and to evaluate to what extent observer experience or patient age may influence observer performance. Material and Methods: A total of 375 patients after hip trauma where radiography was followed by CT or MRI to evaluate a suspected occult hip fracture were collected retrospectively from two imaging centers. After scoring by three observers with varying degrees of radiologic experience, observer variation was assessed by using linear weighted kappa statistics. Results: For radiography, agreements between the three observers were moderate to substantial for intracapsular fractures, with kappa values in the ranges of 0.56-0.66. Kappa values were substantial for extracapsular fractures, in the ranges of 0.69-0.72. With increasing professional experience, fewer fractures were classified as equivocal at radiography. For CT and MRI, observer agreements were similar and almost perfect, with kappa values in the ranges of 0.85-0.97 and 0.93-0.97. Conclusion: There were almost perfect observer agreements for CT and MRI in diagnosing non-displaced, occult hip fractures. Observer agreements for radiography were moderate to substantial, and observer experience influenced agreement only at radiography.