Abstract
BACKGROUND: Costochondral junction fractures are considered specific for child abuse and typically heal with a deformed costochondral junction. OBJECTIVE: To evaluate the types, location, and incidence of costochondral junction variations that can mimic fractures. MATERIALS AND METHODS: A 15-year retrospective study was conducted on children under 2 years of age who underwent chest and abdominal computerized tomography (CT) scans for pneumonia, fever, congenital lung disease, pain, or appendicitis. Randomized selection included 120 chest and 120 abdominal CT scans. Demographic and clinical information was obtained from medical record reviews. Two pediatric radiologists independently reviewed the studies and indicated the presence and location of costochondral junction variation patterns (spurs), and fissure, horizontal lucency, corner, or bucket handle as identified on two consecutive slices on axial views. Disagreements were resolved by a third radiologist. We excluded patients with underlying medical conditions that could affect the skeleton and studies with motion artifacts. A t-test was used to evaluate the relationships between age, CT slice thickness, and the diagnosis of costochondral junction variations. Kappa statistics were used to evaluate agreement. RESULTS: A total of 123 children were excluded due to motion artifacts (n = 30), trauma (n = 31), being evaluated for child abuse (n = 3), slice thickness of 5 mm (n = 1), and underlying medical conditions (n = 58). The final group included 117 children (73 males and 44 females) with an average age of 1 year; 64 had chest and 53 abdominal CT scans. Agreement was fair (kappa = 0.29) at the patient level and poor at the rib level (kappa = 0-0.64). The final number of variations, after resolving disagreements with a third radiologist, was 46 of costochondral junction variations in 19 children (16.2%, 19/117); all were costochondral junction spurs at the levels of the second to eighth ribs. Costochondral junction variations were significantly more common in younger children (average 0.7 ± 0.6 years vs. 1.1 ± 0.6 years, P = 0.024) and when there was thinner CT slice thickness (average 1.6 ± 1.4 mm vs. 2.5 ± 1.5 mm, P = 0.041). CONCLUSION: Costochondral junction variations were identified in 16.2% of children under 2 years of age, and some may mimic healing costochondral junction fractures. There was only fair agreement between radiologists.