The accuracy of navigated vs. robot-assisted pedicle screw placement in spine surgery: a retrospective single-centre comparative study.

STUDY DESIGN: Retrospective comparative cohort study. INTRODUCTION: Accurate pedicle screw placement is critical for spinal stability and for preventing neurological or vascular injury. This study compared screw-placement accuracy and safety between robot-assisted and navigation-guided pedicle screw implantation and explored differences between intraoperative 3D-fluoroscopy and preoperative CT-based surface-matching registration. METHODS: We assessed all consecutive patients who underwent robot-assisted or navigation-guided pedicle screw placement at our department between 2018 and 2022. For robot-assisted placement, screw trajectories were planned on preoperative CT, and intraoperative patient registration was performed using two radiographs fused with a preoperative CT. Navigation-guided placement used either intraoperative 3D-fluoroscopy registration in the final prone position or CT-based surface matching using preoperative supine CT images. Screw positions were evaluated on postoperative CT using a standardised ordinal grading scale from 0 to 4, with Grades 3-4 defined as clinically relevant malpositions. We analysed accuracy using generalised estimating equations with patient-level clustering and adjustment for covariates. We also assessed operative time, radiation exposure, complications, and revision rates. RESULTS: We analysed 1,718 pedicle screws from 268 patients: 728 were robot-assisted and 990 navigation-guided, including 579 using 3D-fluoroscopy and 411 using CT-based surface matching. Mean screw grades were 0.956 for robot-assisted placement, 0.864 for navigation overall, 0.807 for 3D-fluoroscopy, and 0.944 for surface matching. After accounting for patient clustering and adjustment, screw grade did not differ significantly between navigation-guided and robot-assisted placement (aOR 0.87, 95% CI 0.50-1.53; p = 0.632). Grade 3-4 malpositions occurred in 11.7% of robot-assisted and 9.4% of navigation-guided screws, with no significant difference after adjustment. Radiation exposure and fluoroscopy time differed significantly between groups (p < 0.001), with the lowest median values for robot-assisted placement, followed by surface matching and 3D-fluoroscopy. Registration failure occurred in 12/118 (10.2%) robot-assisted patients compared with 2/150 (1.3%) navigation-guided patients. One new postoperative neurological deficit (foot drop) occurred in the robot-assisted group and none in the navigation group. CONCLUSIONS: No significant overall difference in screw-placement accuracy was demonstrated between the two platforms (Robot vs. navigation) or registration modalities in this heterogeneous real-world cohort; however, they showed different operational profiles, particularly concerning intraoperative radiation exposure, fluoroscopy time, and registration failure.