Introduction: The aim of this study was to assess the learning curve of spinal internal fixation with pedicle screws of a spinal fellow (AG) with no previous experience with the technique, during a 2-year fellowship at a referral spinal surgery centre in Australia.

Methods: Patients treated with pedicle screw (PS) placement by the spinal fellow under the supervision of the attending orthopaedic or neurosurgery consultant were included in this study. Postoperative plain x-rays and, in some cases, computed tomography scans (CT scans) were obtained. PS position was assessed by two blinded independent reviewers, one radiologist (observer 1) and one spinal surgeon (observer 2), using a grading scale. PS placed by the attending consultants were included in the evaluation as distractors. The screws were classified using a grading scale in 3 groups: Correct, Border-line, and Incorrect. After assessment, the PS were rearranged in groups of 40 screws, in chronological order, for comparison and assessment of the progress in the learning curve.

Results: 94 patients underwent internal fixation of the spine from upper thoracic to the sacral region with PS (584 screws in total) between February 2006 and December 2007. Eight cases (40 screws) were excluded because of lack of image studies or severe spinal deformities. Among the 544 screws under evaluation, 320 (58.8%) were performed by the spinal surgery fellow, 187 (34.4%) by the attending consultant and 37 (6.8%) by orthopaedic and neurosurgery registrars, the latter evaluated but excluded for the statistical analysis. The overall precision for the 507 screws analysed was 84.2% according to observer 1 and 77.9 % according to observer 2. When the analysis was narrowed down to the 320 screws done by the fellow, this precision increased to 84.7% for observer 1 and decreased to 76.6% according to observer 2. A learning curve was created consisting of 8 groups of 40 screws in chronological order. There was a statistical significance (p< 0.05) in the rate of Incorrect and Border-line PS when comparing the first 100 PS with the rest of the series. None of the patients (included and excluded) developed neurological complications because of the misplaced PS.

Discussion: The findings reveal a learning curve of PS placement. In this series, the inflexion point in the learning curve for this technique was between 80 and 120 screws, which in the present series represented the spinal fellow intervention in 20 to 30 cases. After approximately 150 PS no significant changes can be observed: the learning curve remains stable, with a constant decreasing trend. With appropriate expert supervision the fellowship training system is a safe and appropriate method to learn this technique.