Fracture detection using a low-dose computed tomography paediatric bone fracture phantom: A multi-reader study of radiographers

Introduction: Low-dose computed tomography (CT) has been proposed as an alternative approach to investigate suspected physical abuse (SPA). This is an important consideration for children under 1 year of age due to the inconvenience of initial and follow-up imaging in addition to CT of the head. Results from a pilot study using a bone fracture paediatric phantom suggested that fracture detection was inconsistent using low-dose CT. The aim of this study was to evaluate whether those findings were consistent for a larger cohort of observers. Methods: Nine CT datasets were reviewed by 18 radiographers who marked the fractures using specialist image viewing software. Fractures detected were compared to the known fractures in the commercial paediatric bone fracture phantom. Sensitivity, specificity, positive and negative predictive values (PPV, NPV), with bootstrapped 95 % confidence interval (CI), were calculated. Results: None of the observers detected all seven fractures, with a maximum of five identified by four observers. Bootstrapping predicted that the most commonly detected fractures were of the skull with a sensitivity of 88 %, specificity of 67 %, 95 % CI for PPV ranging from 96 to 99 % and NPV of 15–44 % and the tibia and fibula with a sensitivity of 70 %, specificity of 83 %, 95 % CI for PPV ranging from 93 to 99 % and NPV of 16–45 %. Conclusion: Radiographers had overall sensitivity of 60 % for fracture detection on this phantom, unaffected by radiation dose, but with large variability between observers. Therefore, more research is required to identify the factors that may have influenced this, such as experience, fracture or phantom-based effect. Implications for practice: There is currently no evidence to support a change in current clinical practice which is the radiographic skeletal survey series to investigate SPA in children.