Validation of a new method for the radiological measurement of rod curvature in patients with spine deformity.

Abstract

Purpose: The relationship between rod curvature and postoperative radiographic results is a debated topic. One of the reasons of the heterogeneity of the observed results might reside in the lack of a validated and widely employed method to measure the curvature of the rods. Aim of this study was to present and validate a novel method for rod measurement, which is based on routine X-rays and utilizes a regression algorithm that limits manual measurements and the related errors.

Methods: Data from 20 adolescent idiopathic scoliosis/Scheuermann kyphosis (AIS/SK) patients and 35 adult spine deformity (ASD) patients for analysis, with 112 rods in total. An orthogonal reference grid was overlaid on the lateral X-ray; seven points were then marked along each rod and their coordinates recorded in a table. Using these coordinates, a third-order polynomial regression was applied to obtain the rod curvature equation (correlation coefficients > 0.97). Three observers (one surgeon, one experienced and one inexperienced observer) independently applied the developed method to measure the rod angulation of the included patients and performed the measurements twice. The reliability of the method was evaluated in terms of intraclass correlation coefficient (ICC), Bland-Altmann plot and 2S_R.

Results: The intra-observer ICCs for all measurements exceed 0.85, indicating an excellent correlation. For the AIS/SK group, the surgeon showed a slightly lower reliability compared to the other two evaluators (0.93 vs 0.98 and 0.98). However, the surgeon showed a higher reliability in measurements of the rods at the lumbar level, both for L1-S1 and L4-S1 (0.98 vs 0.96 and 0.89; 0.97 vs. 0.85 and 0.91, respectively). The variability also showed excellent results, with a mean variability ranging from 1.09° to 3.76°. The inter-observer ICCs for the three measurement groups showed an excellent reliability for the AIS/SK group (0.98). The reliability was slightly lower but still excellent for the lumbar measurements in ASD patients at L1-S1 (0.89) and L4-S1 (0.83). The results of the 2S_R for each measured segment were 4.4° for T5-T11, 5.4° for L1-S1 and 5.5° for L4-S1.

Conclusion: The described method represents a reliable and reproducible way to measure rod curvature. This method is based on routine X-rays and utilizes a regression algorithm that limits manual measurements and the related errors.