Accuracy of Ultrasound in the Diagnosis of Classic Metaphyseal Lesions Using Radiographs as the Gold Standard

BACKGROUND: Diagnosis of classic metaphyseal lesions (CMLs) in children suspected for child abuse can be challenging. Ultrasound (US) can potentially help diagnose CMLs. However, its accuracy is unknown. OBJECTIVE: To evaluate the accuracy of US in the diagnosis of CMLs using skeletal survey reports as the gold standard. MATERIALS AND METHODS: US of the metaphysis was performed in three patient groups age <1 year. Informed consent was obtained for patients scheduled for renal US (Group 1) and for patients scheduled for skeletal surveys for possible child abuse (Group 2). Targeted US was also performed in selected patients to evaluate for possible CML suspected on radiographs (Group 3). In Groups 1 and 2, US was performed of both distal femurs, and of either the right or left proximal and distal tibia. Two radiologists (Rad1 and Rad2) independently reviewed the US studies, blinded to history and other imaging. US sensitivity and specificity were calculated using the following gold standards: CML definitely seen on skeletal survey (positive), CML definitely not seen on skeletal survey or part of renal US group (negative). Cases where the skeletal survey was indeterminate for CML were excluded. Kappa statistics were used to evaluate interobserver variability. RESULTS: Two hundred forty-one metaphyseal sites were evaluated by US in 63 children (mean age: 5 months; 33 males); 34 had skeletal surveys and 29 had renal US. Kappa for the presence of CML was 0.70 with 95.7% agreement. US sensitivity was 55.0% and 63.2% and the specificity was 97.7% and 96.7% for Rad1 and Rad2, respectively. CONCLUSION: US has low sensitivity and high specificity in CML diagnosis. Thus, negative US does not exclude CML, but when the radiographs are equivocal, positive US can help substantiate the diagnosis.