A comparative analysis of 4 curvature measurement methods in early-onset scoliosis.

STUDY DESIGN: Observational study with 3 examiners.

OBJECTIVE: The aim of this study was to enhance the reproducibility and reliability of coronal curvature measurements in early-onset scoliosis.

SUMMARY OF BACKGROUND DATA: Previous reports show high variability of the Cobb method, especially on the measurement of the immature spine.

METHODS: A total of 115 whole-spine posteroanterior radiographs were collected to compare the reliability of the Cobb, lateral tangent, pedicle, and centroid methods in early-onset scoliosis. Radiographs were measured twice by each of the 3 examiners using the 4 measurement methods. Statistical analysis was performed to determine the inter- and intraobserver reliability.

RESULTS: In this study, total inter- and intraobserver inter- and intraclass correlation coefficients (ICCs) in 115 radiographs were excellent in all methods (ICCs >0.961). However, mean absolute differences (MADs) in the lateral tangent method were less than 3.78°, which was higher than other methods (MADs <2.95°). In analysis of different severity groups (<15°, 15°-30°, and >30°), total inter- and intraobserver ICCs gradually increased with increasing the severity of the deformity, whereas MADs of each severity group were similar despite their increased measurement scale. Particularly, interobserver ICCs and MADs of lateral tangent method were more than 0.474 and less than 3.76° with poor reliability, which showed high variability in the less deformed spine group (<15°). However, intraobserver ICCs and MADs of pedicle method were more than 0.853 and less than 2.61°, interobserver ICCs and MADs were more than 0.729 and less than 2.86° with excellent reliability, which showed constantly high reliability regardless of coronal curvature severity.

CONCLUSION: In this study, the pedicle method showed constantly higher ICCs and lower MAD values in the early-onset scoliosis regardless of severity. However, the other 3 methods showed lower ICCs and higher MAD values, which showed lowest reliability in the lateral tangent method. For improved treatment of early-onset scoliosis, we recommend the pedicle method for measuring curvature regardless of severity.