Reliability and Reproducibility of Lenke's New Classification System for Idiopathic Scoliosis

Purpose: A new comprehensive classification system of adolescent idiopathic scoliosis has recently been developed by Lenke et al. This system has been introduced with emphasis on sugical planning. For this purpose a high reliability and reproducibility is crucial. This study was designed to determine the intraobserver reliability and the interobserver reliability of Lenke’s new classification system of idiopathic scoliosis.

Methods: Preoperative coronal and sagittal radiographs, as well as side bending films of 51 consecutive patients with idiopathic scoliosis were labeled with the Cobb angles. The center sacral vertical line was marked. The assignment to a curve type (1-6), a lumbar spine modifier (A, B, C), and to the sagittal thoracic modifier (-,N,+) as recently described by Lenke et al. was evaluated independently by five observers. Assignment of the curves was repeated three weeks later with the curves presented in a different order. Kappa coefficients were used to determine the interobserver reliability and the intraobserver reliability.

Results: All 5 reviewers agreed on the overall classification in 21 (41%) of all 51 patients. A mean Kappa value of 0.62 was determined for the interobserver reliability and a mean Kappa value of 0.73 was found for the intraobserver reliability. Disagreement was considered when at least two observers classified a different curve type, or a different lumbar spine modifier, or a different sagittal thoracic modifier. There were overall 56 disagreements when viewing trial 1 and 2 were averaged. Disagreements were especially detected for judging the upper thoracic curve structural or non structural and for assigning a lumbar spine modifier.

Discussion: Disagreement arose from judging the upper thoracic curve. Lenke defined the upper thoracic curve structural, if side bending residual Cobb is 25° or more, regardless of a positive tilt. We are used to evaluate T1 for positive tilt and the shoulder balance. Thus, the new system may interfere to some amount with our older evaluation pattern. Disagreement occured also when classifying lumbar spine modifier B and C. Both issues are clinically important because structural upper thoracic curves and modifier C have to be included in the arthrodeses.

Conclusion: The classification system of Lenke et al. is more reliable than the older King classification, but proper classification of high thoracic and lumbar curves seems to be still difficult.