Cervical spine trauma radiography: comparison of general and musculoskeletal radiologists, with emphasis on number of views

Abstract

We investigated the number of views necessary to evaluate trauma patients for suspected cervical spine fractures, comparing general and musculoskeletal radiologists. At separate sittings, independently and blindly, two musculoskeletal (MSKR) and two general (GR) radiologists evaluated sets of cervical spine images on a 1- to 5-point confidence scale for the presence of fracture in 68 trauma patients. First they evaluated anteroposterior and lateral views, then these plus an odontoid view, and lastly these three plus bilateral oblique radiographs. Overall, MSKR were slightly more accurate than GR (77.9 % vs. 74 %, P < 0.05). With two views, GR were more accurate (75.7 % vs.72.1 %); with three and five views, MSKR were more accurate (80.1 % vs.73.5 % and 81.6 % vs. 72.8 %). Confidence scores for excluding fractures increased with number of views (1.9, 1.6, 1.4, P < 0.0001); however, confidence scores for diagnosing fractures were affected neither by specialty (MSK = 3.8, GR = 3.9) nor by number of views (3.7, 3.9, 4.0). With more views, MSKR were more accurate and confident in excluding suspected cervical spine fractures than GR.