The reproducibility comparison of two intervertebral translation measurements in cervical flexion-extension

The abnormal translations between vertebrae in the sagittal plane are important clues to spinal dysfunction or instability. Several studies have reported significant variability in their translation measurements with no analysis of data reproducibility.

We sought to determine the intra- and interobserver reproducibility of the computer-assisted geometric midplanes and rotation matrix methods in the measurements of intervertebral translations at different motion ranges of cervical flexion-extension in asymptomatic subjects and disc-herniated patients.

A blind, repeated-measure design was applied to determine the reproducibility for intervertebral translation measurements.

A total of 608 videofluoroscopic image sequences from the different motion ranges of cervical flexion and extension in 38 asymptomatic subjects and 38 disc-herniated patients were digitized for further analysis.

The intra- and interobserver reproducibility on measuring the sequential translations were in the acceptable range for geometric midplanes method (average intraclass correlation coefficients [ICCs], 0.860 and 0.806; mean absolute difference [MAD] 0.19 and 0.33 mm) and rotation matrix method (average ICCs, 0.807 and 0.735; MAD, 0.35 and 0.42 mm). There was significantly better reproducibility on the measurements of intervertebral translation for the geometric midplanes method than those of rotation matrix method (p=.001-.040). The absolute mean differences of the translation measurements between two image protocols averaged 11.2% and 10.8% for the asymptomatic subjects and disc-herniated patients, respectively.

Based on these results, both methods demonstrated acceptable reproducibility on the intervertebral translation measurements. The geometric midplanes method involving an averaging effect on the placements of vertebral landmarks and closer to center of rotation might reduce the errors in translation estimations. The rotation matrix protocol simultaneously illustrated horizontal and vertical translation motion despite greater digitizing and/or measurement errors.