Digital Digital Dynamic radiographic criteria for detecting pseudarthrosis Dynamic radiographic criteria for detecting pseudarthrosis following anterior cervical arthrodesis following anterior cervical arthrodesis

Background: While interspinous motion analysis is commonly used to determine the status of an anterior cervical fusion, the accuracy of this technique is unclear. We believed that three questions needed to be answered. What degree of image magnification is ideal? How much motion should be considered ‘‘adequate’’ for making dynamic radiographs? What is the optimal amount of interspinous motion for detecting pseudarthrosis? Methods: We performed a retrospective study of 125 patients (109 fused segments and 153 pseudarthrotic segments) who had undergone reexploration with confirmation of fusion status. Interspinous motion at each operatively treated level and one superjacent level was measured by two independent investigators twice. Reliabilities of interspinous motion analysis at different magnification rates (25%, 100%, 150%, and 200%) were evaluated for fifty randomly selected segments to determine the optimal magnification, which we used for the remainder of the measurements. Fusion status was also determined on computed tomography (CT) by two other raters. We compared the intraoperative findings with those based on dynamic radiographs (with use of cutoff values of 1 and 2 mm of interspinous motion as the indication of pseudarthrosis) and CT. Results: On radiographs, both 150% and 200% magnification yielded higher interobserver and intraobserver reliabilities compared with 25% and 100% magnification, and the reliabilities at 150% and 200% were similar to each other, so subsequent measurements were made at 150%. The cutoff value of interspinous motion for detecting pseudarthrosis was 0.9 mm as determined with receiver operating characteristic curve analysis. Compared with CT, measurements were abnormally distributed, intraobserver and interobserver reliabilities of the agreement with regard to eachmeasurement were assessed with use ofSpearman correlation for interspinous motion measurements and with use of Cohen kappa statistics for CTevaluation. Additionally, we used a Bland-Altman plot analysis of inter-individual difference against the average of the two observers for interspinous motion. Two limits of agreement (mean and ± 1.96 standard deviation [SD]) plots were combined on the graph. Null hypotheses of no difference were rejected if p values were <0.05. study.