Reliability analysis for manual adolescent idiopathic scoliosis measurements

Study Design. Manual radiographic measurement analysis. Objectives. To determine the intraobserver and interobserver reliability of numerous radiographic process measures used in the assessment of adolescent idiopathic scoliosis. Summary of Background Data. Analysis of scoliosis requires a thorough radiographic evaluation to completely assess the deformity. Numerous radiographic process measures have been studied extensively and used for outcomes assessment and thus become the de facto standard of care. However, many of these measures have not been evaluated to determine the reliability and reproducibility. Validation of radiographic process measures is necessary to compare these measures with patient-focused outcome measures, as well as to permit valid comparison of different surgical techniques. Methods. Thirty complete sets of long-cassette scoliosis radiographs (anteroposterior [AP], lateral and side-bending preoperative and AP, and lateral postoperative) were analyzed by three independent experienced observers on two separate occasions. Coronal image measures included the coronal Cobb angles, side-bending Cobb, apical vertebral translation, coronal balance, T1 tilt, lowest instrumented vertebrae (LIV) tilt, angulation of the disc below the LIV, apical vertebral rotation (Nash-Moe), and Risser sign; sagittal measures included T2-T5, T5-T12, T2-T12, T10-L2, T12-S1, and sagittal balance. Intraobserver and interobserver reliability for each measure was then assessed. Results. The vast majority of the radiographic process measures assessed demonstrated good to excellent or excellent intraobserver and interobserver reliability. However, the angulation of the disc below the LIV demonstrated only fair interobserver reliability for postoperative measurements (rho = 0.59). Likewise, Risser grade measurements reflected good intraobserver (0.81-0.99) but only fair interobserver reliability (0.60-0.70). Apical vertebral rotation assessed by the technique of Nash and Moe produced good intraobserver reliability before surgery (0.74-0.85) but only fair reliability after surgery (0.50-0.85). The interobserver reliability for apical Nash-Moe rotation was fair to poor (0.53-0.59). For T2-T5 regional kyphosis, intraobserver (0.22-0.83) and interobserver (0.33-0.47) reliability was generally poor. Overall, the reliability of postoperative measurements tended to be decreased relative to preoperative values, likely due to instrumentation overlying radiographic landmarks. Conclusions. Most of the radiographic process measures evaluated in this study demonstrated good or excellent reliability. The reliability of measuring the angulation of the disc below the LIV, the apical Nash-Moe rotation, and Risser grading was decreased relative to other measures. The reliability of measuring T2-T5 regional kyphosis was disappointing and poor. With regards to the other 13 measures assessed, our findings support the use of these process measures obtained by experienced deformity surgeons via manual measurement for routine clinical and academic purposes.