Interrater reliability of roentgenological evaluation of the lumbar spine in lateral bending.

Interrater reliability is described for the evaluation of lumbar motion and motion coupling patterns in lateral bending. Upright neutral and lateral bending lumbar films of 58 student volunteers were marked by three examiners. Segmental tilt, segmental rotation and motion categories were computed. The performance of the examiners was evaluated using the Kappa statistic and a comparison of rater discrepancy to the experimental measurement error. Although all three raters were proficient within the context of calculated measurement error, good reliability was not universally demonstrated. Concordance for rotation was moderate to good, while agreement for lateral bending was of questionable interpretation. Reliability for the segmental motion categories was good for L1-L4, but unacceptable at L5. The poor reliability at L5 resulted in unacceptable concordance for the lumbar global motion category for three raters. The use of lateral bending radiographs for categorization of the lumbar spine in clinical practice is questioned.