P0994 - Utility of Skull Radiographs in Infants Undergoing 3D Head CT During Evaluation for Child Abuse

Background
Clinical practice guidelines from the American Academy of Pediatrics and American College of Radiology recommend performing both head CT and skeletal survey including AP and lateral skull radiographs (SR) when evaluating infants for abuse. The increased sensitivity of 3-dimensional helical CT (3DCT) may obviate the need to perform both tests. We therefore compared the accuracy of 3DCT and SR for diagnosing skull fractures in infants to determine whether both tests are necessary.

Methods
We identified children <12 months evaluated for abuse undergoing 3DCT and SR between January 2017 and December 2018. 3DCT and SR images were de-identified, randomized, and blindly read by 2 pediatric radiologists who recorded the presence, location, and type of fracture using a standardized pro forma. Discrepancies were resolved by consensus. Inter-rater reliability (IRR) was calculated using Cohen’s kappa statistic (k). Diagnostic accuracy was compared using McNemar’s test.

Results
158 infants with a median age of 4.7 months (IQR 2.4-7.4 months) underwent 3DCT and SR. Consensus reading identified 46 fractures (29.1%) on 3DCT and 40 fractures (25.3%) on SR. IRR was higher for 3DCT (k=0.95) than for SR (k=0.65). 11 fractures were identified on 3DCT but not SR, including 9 parietal (81.8%) and 2 occipital (18.2%). 5 fractures were identified on SR but not 3DCT, including 4 parietal (80.0%) and 1 occipital (20.0%). There was no difference in the diagnostic accuracy of 3DCT and SR (χ2=1.56, p=0.211).

Conclusions
There is no difference in the accuracy of 3DCT and SR for detecting skull fractures in infants evaluated for abuse. Because 3DCT has better IRR and evaluates for both bony and intracranial injuries it is superior to SR. This suggests that SR can be eliminated from skeletal surveys if a 3DCT is performed, thereby reducing radiation exposure without compromising diagnostic accuracy.