Reliability of lateral bending and axial rotation with validity of a new method to determine axial rotation on anteroposterior cervical radiographs

Abstract

Objective: To investigate the reliability of a new radiographic measurement of axial rotation and lateral bending on anterior-posterior cervical views by using a computer and sonic digitizer. Design: A blind, repeated-measure design was used. Anteroposterior cervicothoracic radiographs were presented to each of 3 examiners in random order. Each film was digitized, and 1 week later the films were randomized for a second run. Setting: Private, primary-care chiropractic clinic. Main Outcome Measures: The interclass and intraclass correlation coef cients (ICC) for intraexaminer and interexaminer reliability were calculated from measurements on radiographs for determining axial rotations (Ry) and lateral bending (Rz) of C3 to T3. Results: When the new axial rotation method was applied to small rotations of a C3 plastic model, the average error was less than 1°. For the calculations of axial rotation (Ry), the ICC values were in the good to excellent range. For axial rotation, the intraclass correlation coefcients were ICCs ≥0.78, and the interclass correlation coefcients were ICCs ≥0.67. For lateral flexions (Rz) of C3 to T3, all intraclass and interclass correlation coefficients were in the excellent range (ICCs > 0.87). Conclusions: Methods of calculating axial rotations in the spine have been reported for large angles (5° to 30°) but not for smaller angles. A new method for determining axial rotations of the cervical segments on AP views, based on the chord across the arc displaced by the spinous-lamina junction, had reliability (ICC values) in the good to excellent range. Compared with measured rotations of a C3 model (-5° to +5°), the new method had an average error of less than 1° and approximately 11.5%. The reliability for the axial rotation measurements was in the good to excellent range, and the lateral bending measurements were all in the excellent range.