Reliability analysis of Cobb measurement in degenerative lumbar scoliosis using endplate versus pedicle as bony landmarks

Degenerative changes of endplates in older patients and tilting of vertebral body in lumbosacral lordosis could make an accurate identification of endplates for the Cobb measurement difficult. Pedicles have been proposed as alternative landmarks because they are usually better visualized, and offer similar clinical validity to the endplates. The objective of this study was to investigate the reliability of the pedicle method of Cobb measurement in degenerative lumbar scoliosis and compare it with the traditional endplate method.

Two hundred and eighty-four radiographs of degenerative lumbar scoliosis were evaluated. The radiographs were classified into groups based on the patient's age (< 60 years, 60 to 80 years, and > 80 years), level of lower end vertebra (LEV) (LEV at L5, and LEV at or above L4), and curve severity (< 20°, 20° to 40°, and > 40°). Three observers independently measured the radiographs using the endplate and pedicle methods twice with an interval of 1 week. The intra- and interobserver reliabilities were calculated using intraclass correlation coefficients (ICC).

The intra- and interobserver ICC values were better for all observers in the > 80 years age group using the pedicle method. The intraobserver ICC values of pedicle method were also better in the LEV at L5 group, and the interobserver ICC values showed a slightly better consistency with the pedicle method. For patients with > 40° curves, the intraobserver ICC values for all observers as well as interobserver ICC values were better using the endplate method.

The reliabilities of the endplate and pedicle methods for degenerative lumbar scoliosis were both excellent. The pedicle method might be better in older patients (> 80 years) and those with LEV at L5; while the endplate method could have some strength in severe cases (> 40°).