Evaluation of accuracy of plain radiography in determining the Risser stage and identification of common sources of errors

Abstract

Background: Risser's sign is an established radiological marker for predicting growth potential in adolescent idiopathic scoliosis, but the accuracy of Risser's staging has been debated. This research was designed to evaluate the accuracy of Risser's staging and to identify causes of errors in Risser's staging. Materials and methods: Plain radiographs of 89 adolescent idiopathic scoliosis patients were evaluated for Risser's stage using both the Original and French methods. A three-dimensional computed tomography (3D-CT) was used to evaluate the accuracy of the plain radiographs. Inter-and intra-observer reliability of both methods was assessed on radiographs and 3D-CT images using weighted kappa statistics. The concordance rate for Risser's staging between plain radiographs and 3D-CT images were calculated. The various sources of staging differences between the two imaging methods were noted, grouped, and analyzed to identify common error patterns. Results: Intra-and inter-observer staging reliabilities on plain radiography were 0.91 and 0.94, respectively, using the Original method and 0.91 and 0.92, respectively, using the French method. Intra-and inter-observer reliabilities on 3D-CT were 0.98 and 0.99, respectively, using the Original method and 0.97 and 0.99, respectively, using the French method. Mean concordance rates between plain radiography and 3D-CT were 59.76% and 67.42% using the Original and French methods, respectively. Common sources of error leading to misinterpretation of Risser's staging were miscalculation of apophysis excursion, skip ossification, isolated non-linear ossification, micro-fusion, and pseudo-fusion. Conclusions: Risser's staging by plain radiography is reliable but not accurate. Variations in the iliac apophysis ossification and misinterpretation of apophysis fusion are the main sources of error.