Purpose: To investigate the intra- and inter-observer reliabilities of the newly developed i-Scolioroller for scoliosis screening, and to determine the optimal i-Scolioroller measurement cutoff values for identifying adolescent scoliosis with a Cobb angle ≥ 20°.
Methods: The i-Scolioroller displays the right- and left-side maximum inclination angle (Rmax, Lmax) during the forward bending test (FBT), as well as the angle of trunk inclination (ATI, i.e., whether the Rmax or Lmax is greater). Sum-ATI is defined as the sum of Rmax and Lmax. Intra-class correlation coefficients (ICC) of the ATI and sum-ATI measurements were calculated to analyze the intra- and inter-observer reliabilities for 10 plaster torsos in FBT positions obtained from patients with idiopathic scoliosis. The optimal cutoff values for scoliosis were determined using receiver operating characteristic (ROC) analysis of i-Scolioroller measurements versus Cobb angles obtained from the upright whole-spine radiographs of 112 adolescent outpatients.
Results: The intra-observer ICCs for the ATI/sum-ATI for 3 observers were 0.851/0.856, 0.786/0.900, and 0.772/0.796, respectively, while the corresponding inter-observer ICCs for all participants were 0.733/0.745. On ROC analysis, an ATI of 8° was the optimal cutoff value for scoliosis (sensitivity and specificity: 79.2% and 70.0%, respectively). The optimal cutoff value for sum-ATI was 11° (sensitivity and specificity: 86.1% and 82.5%, respectively). The areas under the ROC curves were 0.859 for ATI and 0.908 for sum-ATI.
Conclusion: The optimal cutoff values for identifying scoliosis using the i-Scolioroller were a combination of 11° for the sum-ATI and 8° for the ATI.
Keywords: Cobb angle; Intra-class correlation coefficients; Receiver operating characteristic curve; i-Scolioroller.