When Will You Succeed Casting Patients with Early-onset Scoliosis? Prospective Evaluation of Predictive Radiographic Parameters

Background: In a recent retrospective study, in cast correction of the major curve correlated with final curve size in patients with early-onset scoliosis treated with casting. We therefore sought to perform a prospective study with controlled methodology to determine if there are parameters associated with reduction of coronal deformity. Methods: A prospective, observational study was conducted between 2014 and 2019 at selected sites willing to comply with a standard radiographic and follow-up protocol. Radiographic data was collected at time points of precast, in traction, initial in-cast, and at minimum 1 year follow-up. Multivariate linear regression models were utilized to control for potential confounders using a stepwise procedure. Twenty-nine patients met inclusion criteria. Results: On multivariate analysis, traction major curve (P=0.043) and initial in-cast (P=0.011) major curve Cobb angles were independently associated with final out of cast major curve Cobb angle. The only factor that was independently associated with failure to cure (<15-degree major curve) was traction major curve Cobb angle (P=0.046). A threshold traction major curve Cobb angle of 20 degrees was found to have good accuracy with 81% sensitivity and 73% specificity (receiver operator curve area: 0.869, P<0.001). A traction major curve Cobb angle over 20 degrees would accurately predict failure of casting treatment to cure scoliosis in 79% of cases. A threshold in-cast major curve Cobb angle of 21 degrees was found to have slightly less accuracy than traction with 69% sensitivity, 82% specificity, and 74% accuracy (receiver operator curve area: 0.830, P=0.004). Conclusions: Radiographic measurements in traction and initially in the cast are predictive of curve size at follow-up for children with early-onset scoliosis treated with casting. The standardization and utility of traction films should be further explored. Level of Evidence: Level II.