Unskilled unawareness and the learning curve in robotic spine surgery.

Background: Robotic assistance for the placement of pedicle screws has been established as a safe technique. Nonetheless rare instances of screw misplacement have been reported.The aim of the present retrospective study is to assess whether experience and time affect the accuracy of screws placed with the help of the SpineAssist™ robot system.

Methods: Postoperative computed tomography (CT) scans of 258 patients requiring thoracolumbar pedicle screw instrumentation from 2008 to 2013 were reviewed. Overall, 13 surgeons performed the surgeries. A pedicle breach of >3 mm was graded as a misplacement. Surgeons were dichotomised into an early and experienced period in increments of five surgeries.

Results: In 258 surgeries, 1,265 pedicle screws were placed with the aid of the robot system. Overall, 1,217 screws (96.2 %) were graded as acceptable. When displayed by surgeon, the development of percent misplacement rates peaked between 5 and 25 surgeries in 12 of 13 surgeons. The overall misplacement rate in the first five surgeries was 2.4 % (6/245). The misplacement rate rose to 6.3 % between 11 and 15 surgeries (10/158; p = 0.20), and reached a significant peak between 16 and 20 surgeries with a rate of 7.1 % (8/112; p = 0.03). Afterwards, misplacement rates declined.

Conclusions: A major peak in screw inaccuracies occurred between cases 10 and 20, and a second, smaller one at about 40 surgeries. One potential explanation could be a transition from decreased supervision (unskilled but aware) to increased confidence of a surgeon (unskilled but unaware) who adopts this new technique prior to mastering it (skilled). We therefore advocate ensuring competent supervision for new surgeons at least during the first 25 procedures of robotic spine surgery to optimise the accuracy of robot-assisted pedicle screws.