Accuracy and Reliability of Drawing Central Sacral Vertical Line on Scoliosis Radiographs in Clinical Practice

Abstract

Study DesignRetrospective cross-sectional assessment of radiographs of adolescent idiopathic scoliosis. ObjectiveTo assess accuracy, reliability, and reproducibility of the central sacral vertical line (CVSL) on scoliosis radiographs and its influence on determining clinically relevant parameters: stable vertebra (SV) and lumbar modifier (LM). Background SummaryThe central sacral vertical line is frequently used on scoliosis radiographs for surgical planning. However, no studies have assessed how accurately and reliably a physician draws CSVL in routine clinical practice. MethodsWe provided 30 digital posteroanterior X-rays of adolescent idiopathic scoliosis to 5 raters (3 fellowship-trained spine surgeons and 2 fellows) at 2 different times (3-week interval) to determine SV and LM. An independent observer then assessed the accuracy of CSVL drawn by the raters. The CSVL was considered accurate when it was drawn vertical from the middle of the S1 vertebra. To avoid conscious bias, we kept raters blinded as to the actual purpose of the study. To avoid technical bias, the study was conducted with research Picture Archiving and Communication Systems (PACS) software and raters used desktop computers similar to those in clinics or operating rooms. ResultsBased on absolute values, the CSVL was drawn on average 2.3 (±1.9) mm away from the center. Based on raw values, the lines were drawn on average 1.0 (±2.8) mm left of the center. We observed fair inter-rater reliability among the 5 raters, with an intra-class correlation of 0.23 (95% confidence interval, −0.33–0.59). Intra-rater reproducibility was moderate across 2 assessments, with an ICC of 0.47 (95% confidence interval, 0.27–0.62). Variation in drawing CSVL from center was not significantly associated with the selection of SV (r=−0.02; p=.78) or LM (r=−0.02; p=.78). ConclusionAlthough, physicians draw CSVL significantly away from the center of the S1 vertebra (mean, 2.3 mm), its influence on determining SV or LM is not affected in routine practice.