Factors Affecting the Accuracy of Pedicle Screw Placement in Robot-Assisted Surgery: A Multicenter Study.

Abstract

Retrospective multicenter. The aim was to investigate the factors involved in, and their relative contributions to, the overall accuracy of robot-assisted pedicle screw placement. Robot-assisted surgery has reportedly resulted in greater accuracy for placement of pedicle screws than conventional methods. There are many potential factors affecting the accuracy of pedicle screws placed with a robot. No study has investigated these factors in a robust way. Radiographic and clinical data of three centers were pooled. Preoperative and postoperative computerized tomographies were obtained by all three centers to assess the accuracy of the placed screws. The primary outcome measured was accuracy of pedicle screws placed with the robot. The authors performed a multivariate regression analysis to determine the significant patient-related and screw-related variables and their relative contribution to the overall accuracy. In addition, an ordinal regression analysis was conducted to investigate the effects of different variables on accuracy of robot-placed screws graded by Gertzbein-Robbins grading system (GRS). The total contribution of all studied variables to overall accuracy variation as measured by offsets between the placed and planned screws was only 18%. Obesity, long constructs, female gender, surgeon, and vertebral levels were among the factors that had small contributions to the different screw offsets. For GRS grades, significant variables were gender (Log odds: 0.62, 95% CI: 0.38-0.85), age (Log odds: 0.02, 95% CI: 0.01-0.03), length of constructs (Log odds: 0.07, 95% CI: 0.02-0.11), screw diameter (Log odds: 0.55, 95% CI: 0.39-0.71), and length of the screws (Log odds: 0.03, 95% CI: 0.01-0.05). However, these variables too, regardless of their significant association with the accuracy of placed screws, had little contribution to overall variability of accuracy itself (only about 7%). The accuracy of screws placed with robotic assistance, as graded by GRS or measured offsets between planned and placed screw trajectories, is minimally affected by different patient-related or screw-related variables due to the robustness of the robotic navigation system used in this study. Level III.