Abusive head trauma in children: radiographs of the skull do not provide additional information in the diagnosis of skull fracture when multiplanar computed tomography with three-dimensional reconstructions is available

Background Some data suggest that CT is more accurate than skull radiographs in diagnosing skull fractures in abusive head trauma. Objectives We investigated whether multiplanar CT with three-dimensional (3-D) reconstructions alone is non-inferior to combination CT/skull radiography for the diagnosis of skull fracture in suspected physical abuse. Materials and methods We identified children who had skull radiographs and concurrent multiplanar CT with 3-D reconstructions obtained during suspected physical abuse imaging between 2010 and 2019, and a fracture diagnosis in the formal report. We included all fracture cases and an equal number of randomly selected non-fracture controls in an anonymised dataset. This dataset was independently reviewed for skull fracture by two paediatric radiologists and one less-experienced trainee using either radiography alone, CT alone or CT/skull radiography. The primary outcome was discordance in diagnosis of skull fracture between CT alone and CT/skull radiography, with a result > 0.03 deemed to indicate inferiority of CT alone. The Fleiss kappa was used to assess interobserver agreement. Results We included 106 children, 53 with and 53 without skull fracture. A single case was discordant between CT alone and CT/skull radiography, resulting in discordance of 0.009, consistent with non-inferiority of CT alone. The sensitivity and specificity of CT alone and CT/skull radiography were 98% and 96–98%, respectively, whereas radiography alone was more inaccurate (81% sensitivity and 96% specificity). Interobserver agreement for all modalities was very high (kappa 0.86–0.95). Conclusion Multiplanar CT with 3-D reconstructions alone is not inferior (and clinically equivalent) to CT/skull radiography for diagnosing skull fracture in suspected physical abuse imaging and was as accurate when reported by a less-experienced trainee. This suggests that skull radiography can be removed from suspected physical abuse imaging guidelines.