Learning curve of 3D fluoroscopy image-guided pedicle screw placement in the thoracolumbar spine

Background context During the past decade, a disproportionate increase of spinal fusion procedures has been observed. Along with this trend, image-guided spine surgery has been experiencing a renaissance in the recent years. A wide range of different navigation systems are available on the market today. However, only few published studies assess the learning curves concerning these new spinal navigation techniques. So far, a study on the learning curve for intraoperative three-dimensional fluoroscopy (3DFL)-navigated pedicle screw (PS) placement is still lacking. Purpose The purpose of the study was to analyze the learning curve for 3DFL-navigated thoracolumbar PS placement. Study design/setting The study design included a prospective case series. Patient sample A cohort of 145 patients were recruited from January 2011 to June 2012. Outcome measures The outcome measures were duration of intraoperative 3D scans, PS placement, PS accuracy on postoperative computed tomography (CT) scans, and PS-related revisions and complications. Methods From the introduction of spinal navigation to our department in January 2011 until June 2012, the learning curve for the duration of intraoperative 3D scan acquisition (navigation or control scan) and placement time per screw, intraoperative screw revisions, screw-related complications, revision surgeries, and PS accuracy on postoperative CT scans were assessed in 145 patients undergoing dorsal navigated instrumentation for 928 PS (736 lumbosacral and 192 thoracic). The observed time span was divided into four intervals. Results of the second, third, and last periods were compared with the first (reference) period, respectively. Results The mean navigation 3D scan time decreased (first and fourth periods) from 15.4±7.8 (range, 4-40) to 8.4±3.3 (3-15) minutes (p<.001). The mean control 3D scan time (after PS placement) decreased from 11.2±4.8 (5-25) to 6.6±3.0 (3-15) minutes (p<.001). The mean PS insertion time decreased from 5.3±2.5 (1-15) to 3.2±2.3 (1-17) minutes (p<.001). The mean proportion of correctly positioned PS (all 928) according to the Gertzbein and Robbins classification grades A and B increased initially from 83.1% (first period) to 95.1% (second period, p=.001), 96.4% (third period, p=.002), and 92.4% (fourth period, p=.049). No learning effect was found with respect to intraoperative screw revisions. There was one revision surgery. Conclusions We could demonstrate significant learning effects for 3DFL-navigated PS placement with regard to intraoperative 3D scan acquisition, PS placement time, and PS accuracy.