Skull radiograph interpretation of children younger than two years: how good are pediatric emergency physicians?

STUDY
OBJECTIVE: We determine pediatric emergency physicians' accuracy in interpreting skull radiographs of children younger than 2 years and determine the characteristics of misidentified skull radiographs.
METHODS: A set of 31 skull radiographs (16 with fractures, 15 normal) was compiled from children younger than 2 years who were evaluated for head trauma in a pediatric emergency department from March 3, 1997, to March 3, 1998. A pediatric radiologist reinterpreted the films and agreed with all of the original readings in the final set. Participants (attending level physicians) were asked to identify the presence, location, and pattern of any fracture. Skull radiograph interpretation was considered radiographically correct if the presence, location, and pattern of fracture were correctly identified and was considered diagnostically correct if the presence of a fracture was recognized.
RESULTS: Twenty-five of 26 eligible pediatric emergency physicians completed the study. The mean of each participant's radiographically correct interpretation was 65%+/-10% (mean+/-SD), and diagnostically correct interpretation was 80%+/-9%. The group's mean sensitivity for diagnostically correct interpretation was 76%+/-15%, and specificity was 84%+/-14%. Shorter fractures were identified correctly less often (63% < or =5 cm versus 93% >5 cm; mean difference 30%; 95% confidence interval 21% to 39%). Diagnostically correct rates did not differ according to age of patient, physician practice location, years in practice, or practice in ordering skull radiographs.
CONCLUSION: Pediatric emergency physicians have limited accuracy in interpreting skull radiographs of children younger than 2 years. Shorter fractures are more commonly misinterpreted.