Comparison of reliability between the Lenke and King classification systems for adolescent idiopathic scoliosis using radiographs that were not premeasured

Abstract

Study design: Multisurgeon comparison of two radiographic scoliosis curve classification systems was performed.

Objective: To determine the reliability of the King and Lenke classifications systems for adolescent idiopathic scoliosis using radiographs that had not been premeasured.

Summary of background data: Recent studies introducing the new Lenke classification system for idiopathic scoliosis have reported reliability improved over that of the King classification system. This newer classification system evaluates three different parameters (curve type, lumbar modifier, and sagittal thoracic modifier) and then combines them. The reliability of both classification systems had been determined using radiographs in which all of the curves had been premeasured (recorded on the radiographs) before review by examiners. However, in a normal clinical situation, spine surgeons need to determine the Cobb angles independently, thus introducing another variable.

Methods: On two separate occasions, four orthopedic surgeons independently evaluated preoperative radiographs (standing posteroanterior, lateral, and two supine side-bending views) of 50 patients with adolescent idiopathic scoliosis. All measurements had been removed on every radiograph before each evaluation. The results were determined by calculating the average percentage of intraobserver and interobserver agreement. Reliability was quantified using kappa statistics.

Results: The King classification demonstrated good intraobserver and fair interobserver reliability. Intraobserver percentage of agreement averaged 83.5% (kappa coefficient, 0.81). Interobserver percentage of agreement averaged 68.0% (kappa coefficient, 0.61). All three parameters of the overall Lenke curve classification demonstrated fair reliability. Intraobserver percentage of agreement averaged 65.0% (kappa coefficient, 0.60). Interobserver percentage of agreement averaged 55.5% (kappa coefficient, 0.50). When the Lenke curve type was examined separately, intraobserver percentage of agreement averaged 81.5% (kappa coefficient, 0.76) and interobserver percentage of agreement averaged 71.5% (kappa coefficient, 0.64). The results for this variable (curve type) were similar to those for the King classification. For the Lenke lumbar modifier, the percentage of agreement and reliability were excellent. For the sagittal thoracic modifier, the percentage of agreement was good, but the kappa values were low because of an extreme imbalance in the grouping of hypokyphotic, normal, and hyperkyphotic spines.

Conclusions: In this study, with each investigator performing the radiographic measurements, the King classification was found to be better than had been reported recently. The Lenke classification system for adolescent idiopathic scoliosis was found to be less reliable than previously reported when the radiographs were premeasured. This was particularly true when all three parameters of this new classification system were combined. This difference in reliability of the Lenke classification between studies can be attributed to the additional variable of determining the Cobb measurements on each of the unmarked radiographs. Although this new classification system has limitations with respect to interobserver and intraobserver reliability, for planning operative treatment, it offers a more comprehensive radiographic evaluation of patients with adolescent idiopathic scoliosis than previous systems.