Background: A fracture classification system enables communication among surgeons and provides guidelines for treatment as well as some estimate of prognosis. Thus, the system should be anatomically meaningful and reliable. The purpose of this study was to assess the interobserver and intraobserver reliability of Letournel's acetabular fracture classification and the effect of computed tomography on its reliability.
Methods: Plain radiographs (anteroposterior and Judet views) and axial computed tomography scans were randomly chosen from an acetabular fracture database, with at least five cases of each fracture type and eight of the most common types. The study involved three groups of three orthopaedic surgeons: (1) surgeons who had studied under Letournel, (2) surgeons who specialized in acetabular fracture surgery, and (3) general trauma surgeons. Each observer read the radiographs twice, and at each session the fractures were classified first on the basis of the radiographs only and then in combination with the computed tomography scan. Observer agreement was then assessed with the unweighted kappa coefficient (?). We also calculated the frequency with which the observers agreed with the diagnosis made intraoperatively by the treating orthopaedic surgeon.
Results: The interobserver reliability without and with computed tomography during the first session was 0.70 and 0.74, respectively, for group 1, 0.71 and 0.69 for group 2, and 0.51 and 0.51 for group 3. The results of the second session were similar. When the two sessions were compared, intraobserver reliability without and with computed tomography was 0.80 and 0.83 for group 1, 0.80 and 0.80 for group 2, and 0.64 and 0.69 for group 3. The overall agreement of the radiographic observation with the fracture pattern observed at surgery was 74%.
Conclusions: Letournel's acetabular classification with use of plain radiographs with or without supplemental computed tomography scans has substantial reliability (? > 0.7) when used by surgeons who have been taught how to interpret the images or by those who treat acetabular fractures on a regular basis. The value of computed tomography scans in the evaluation of acetabular fractures has been well established for the identification of loose bodies and articular impaction; however, they do not appear to be essential for the classification of acetabular fractures.