Predicting scoliosis progression from skeletal maturity: a simplified classification during adolescence.

Background: Both the Tanner-Whitehouse-III RUS score, which is based on the radiographic appearance of the epiphyses of the distal part of the radius, the distal part of the ulna, and small bones of the hand, and the digital skeletal age skeletal maturity scoring system, which is based on just the metacarpals and phalanges, correlate highly with the curve acceleration phase in girls with idiopathic scoliosis. However, these systems require an atlas and access to the scoring system, making their use impractical in a busy clinical setting. We sought to develop a simplified system that would correlate highly with scoliosis behavior but that would also be rapid and reliable for clinical practice.

Methods: A simplified staging system involving the use of the Tanner-Whitehouse-III descriptors was developed. It was tested for intraobserver and interobserver reliability by six individuals on thirty skeletal age radiographs. The system was compared with the timing of the curve acceleration phase in a cohort of twenty-two girls with idiopathic scoliosis.

Results: The average intraobserver unweighted kappa value was 0.88, and the average weighted kappa value was 0.96. The percentage of exact matches between readings for each rater was 89%, and 100% of the differences were within one unit. The average interobserver unweighted kappa value was 0.71, and the average weighted kappa value was 0.89. The percentage of exact matches between two reviewers was 71%, and 97% of the interobserver differences were within one stage or matched. The agreement was highest between the most experienced raters. Interobserver reliability was not improved by the use of a classification-specific atlas. The correlation of the staging system with the curve acceleration phase was 0.91.

Conclusions: The simplified skeletal maturity scoring system is reliable and correlates more strongly with the behavior of idiopathic scoliosis than the Risser sign or Greulich and Pyle skeletal ages do. The system has a modest learning curve but is easily used in a clinical setting and, in conjunction with curve type and magnitude, appears to be strongly prognostic of future scoliosis curve behavior.