Centroid method: an alternative method of determining coronal curvature in scoliosis. A comparative study versus Cobb method in the degenerative spine

Background context: Although the Cobb method is considered the gold standard, the centroid method may offer a reasonable alternative in adult scoliosis because it has better inter- and intraobserver reliabilities in adolescent scoliosis.

Purpose: To compare the reliabilities of the Cobb and the centroid methods for measuring coronal curvature in degenerative scoliosis in older patients.

Study design: Observational study involving three examiners.

Patient sample: Sixty whole spine posteroanterior (PA) radiographs were collected.

Outcome measures: Data were analyzed to determine inter- and intraobserver reliabilities.

Methods: Sixty whole spine PA radiographs of patients older than 60 years were collected to compare the reliabilities of the centroid and the Cobb methods for measuring coronal curvature in degenerative scoliosis. Three examiners using both methods independently measured radiographs twice. Data were analyzed to determine inter- and intraobserver reliabilities.

Results: Intraobserver comparisons of all the 60 radiographs revealed that inter- and intraclass correlation coefficients (ICCs) of the Cobb and the centroid methods were both excellent (greater than 0.979 vs. greater than 0.918), and mean absolute differences (MADs) were similar (less than 1.58 vs. less than 2.02). In interobserver comparisons, ICCs of the Cobb method were higher than that of the centroid method (greater than 0.922 vs. greater than 0.799), and the MADs of the Cobb method were lower than that of the centroid method (less than 2.91 vs. less than 4.84). Comparisons of radiographs subdivided by severity showed that the ICCs of the Cobb and the centroid methods were both excellent (greater than 0.819 vs. greater than 0.801), and their MADs were similar (less than 2.29 vs. less than 2.53) for intraobserver comparisons. Interobserver comparisons showed that ICCs and MADs were dependent on the severity of coronal curvature, and the ICCs of the Cobb method (greater than 0.698) were greater than that of the centroid method (greater than 0.507). Furthermore, MAD values for the Cobb method were lower than that for the centroid method (less than 3.59 vs. less than 6.07). Moreover, these results are contradictory to the previous study, which showed the higher reliability of the centroid method in measures of adolescent scoliosis.

Conclusions: In the present study, the reliability of the centroid method was found to be more susceptible to the severity of disease in older patients, despite its demonstrated greater reliability in adolescent scoliosis. Our findings show that the selective use of these two methods in old and young patients can increase the reliabilities of measurements made.