Index radiographic measurements in the prediction of progression in infantile idiopathic scoliosis: a comparative analysis and description of a novel predictive model.

Abstract

To assess the comparative accuracy of commonly utilised index radiological measurements in the prognosis of infantile idiopathic scoliosis (IIS) and build a parsimonious prognostic model utilising these measurements. This was a retrospective analysis of a UK population of patients with IIS. Index radiological parameters were analysed, and outcome of their condition was determined over long-term follow-up. Comparative accuracy of each radiological measurement category was determined by logistic regression analyses and the corresponding receiver operating characteristic (ROC) curve. A predictive model of IIS progression using these measurements was then created. All three radiological measurement categories were predictive of IIS progression. However, on pairwise comparison of ROC curves and multivariate analysis, the index Cobb angle proved the most significant predictor of curve progression. Using the index Cobb angle only, a predictive model of curve progression achieved an accuracy of 81.18% with a cut-off Cobb angle of 34.5° found to be the optimal threshold to discriminate a progressive from resolving curve. Of the three analysed index radiological parameters commonly used by surgeons in the prognosis of IIS, we found that the Cobb angle is the most accurate predictive measure. Further, neither addition of the RVAD nor convex RVA provided significant further prognostic value in a multivariate model of progression. However, not even the Cobb angle model was accurate in all cases; emphasising caution should be applied when relying on index radiological measurements to predict IIS outcomes.