Agreement Between Manual and Computerized Designation of Neutral Vertebra in Idiopathic Scoliosis

Survey-based cross-sectional study.

To describe interobserver agreement among experienced spine surgeons in choosing neutral vertebra (NV) based on manual measurements from radiographs. Secondarily, to use axial vertebral rotation (AVR) values obtained from low-dose stereoradiography (SR) post-processing software (SterEOS 2D/3D) to separately designate the NV in subject cases and to compare manually derived and software-derived NV designations.

Investigators have previously suggested that parameters such as Lenke classification, stable vertebra level, end vertebra level, and NV level be used to decide on fusion levels in adolescent idiopathic scoliosis (AIS). Studies have revealed suboptimal interobserver reliability in these vertebral designations. SR post-processing software may represent a useful tool for standardizing NV designation.

Thirty-two subjects with idiopathic scoliosis and Lenke 1-4 curves were assessed. Experienced surgeons (range of 7-35 years in practice) assigned NV based on preoperative radiographs. Interobserver reliability was quantified using the Fleiss Kappa statistic. Surgeon responses were compared with NV designations made using AVR values provided by SR postprocessing software. Agreement between these values was quantified using percentage agreement.

Surgeons exhibited moderate agreement in choosing NV based on radiographs (Kappa 0.444). Surgeon responses agreed with the SR-derived NV in 26.9% of cases, lay within 1 level in 82.1% of cases, and lay within 2 levels in 97.5% of cases. Surgeons were more likely to choose distal to the SR NV rather than proximal.

Variability in instrumented level selection and outcomes in idiopathic scoliosis may be partially related to inconsistency in selection of the NV. The use of SR post-processing software may provide a more reliable method for choosing NV.

Level II.