West O C, Anbari M M, Pilgram T K, Wilson A J
Mallinckrodt Institute of Radiology, Washington University School of Medicine, St Louis, MO 63110, USA.
Radiology. 1997 Sep;204(3):819-23. doi: 10.1148/radiology.204.3.9280266.
To compare the diagnostic accuracy of single-view and three-view cervical spine radiographs in acute trauma.
Radiographs of 97 patients with acute cervical spine fractures were matched with those of 92 proved normal cases. Fourteen radiologists, including three senior attending radiologists each with more than 20 years of post-residency experience, four junior attending radiologists each with less than 5 years of practice experience, three senior radiology residents in their 3rd or 4th year of radiology residency, and four junior radiology residents at the end of their 1st year of radiology training, interpreted each case twice: once as a lateral-only study and again as a three-view study. Multireader-multicase receiver operating characteristic analysis was performed. Difficult cases were reviewed for trends in the errors.
Eight of 14 readers detected fractures better with the three-view series. Among junior attending radiologists, the differences were statistically significant. Most improvements were in fractures of the dens and fractures and unilateral dislocations of the articular mass. Overall, sensitivity increased from 81.8% to 83.3%. The reliability of fracture classification improved. Less experienced readers performed better with the three-view series.
A three-view screening radiographic series will allow most readers to detect a few more fractures than a single-view series. The improvements occur primarily in a subset of difficult-to-diagnose injuries.
