Reliability of lumbar lordosis measurement in patients with spondylolisthesis: a case-control study comparing the Cobb, centroid, and posterior tangent methods.

STUDY DESIGN

A radiologic analytical study.

OBJECTIVE

To compare the reliability of lumbar lordosis (LL) measurement in cases of spondylolisthesis using the Cobb, vertebral centroid, and posterior tangent methods.

SUMMARY OF BACKGROUND DATA

Interobserver and intraobserver reliabilities of the several different types of lumbar curvature analysis have been reported using various methods for measurement; however, in patients with spondylolisthesis, it has not been studied till yet.

METHODS

A radiologic analytical study was performed in 50 patients who had spondylolisthesis (28 at L4-L5, 22 at L5-S1) with lysis in the lumbar spine. All patients had instability of more than 10 degrees on the lumbar spine flexion and extension radiogram. A total of 26 patients who presented for backache without spondylolisthesis were considered as control group. Three observers measured the global and segmental angle for LL using various methods for measuring the LL using the Cobb, vertebral centroid, and Harrison's posterior tangent methods. All measurements were done by each of the observers on the same computer by keeping at least 2 weeks of interval between 2 sets of measurements. The interobserver and intraobserver reliability was calculated among all the 3 observers' 2 sets of measurements by using the inter- and intraclass correlation coefficient (ICC) test on SPSS program.

RESULTS

Five global and 17 segmental angles measured overall. Results showed an excellent ICC (>0.7) while measuring the global LL using any method, the Cobb (T12-S1 or L1-L5), vertebral centroid, and Harrison's posterior tangent (T12-S1 or L1-L5) methods, in both subject and control groups. Similarly measuring the segmental angles, it exhibited excellent intraclass correlation (ICC, >0.7) coefficient by using all 3 methods for all segmental angles in both groups. However, interclass correlation coefficient was excellent (ICCs, >0.7) in both global as well as segmental angles for vertebral centroid and posterior tangent methods in subject and control groups, but it was poor or fair for the Cobb methods in segmental angles and good or excellent in global angles. Measuring the segmental angles, standard error of mean (SEM) was <2 degrees at all the levels by all methods in both subject and control group except the L4-L5-S1 level by centroid method in subject group where SEM >2 degrees was found.

CONCLUSION

Posterior tangent method should be used for the global and segmental angle analysis for the LL in cases with spinal instability because of (a) higher correlation coefficient for segmental angle measurements; (b) lower SEM at the instability level than the centroid method despite similar correlation coefficients; and (c) similar to the engineering analysis.