Navigation versus experience: providing training in accurate lumbar pedicle screw positioning

Pedicle screw (PS) placement, which is considered complex and technically demanding, is a widely used technique for surgical correction of deformities, degeneration, infection, malignancy, and trauma of the spine. Adequate PS placement presents an important predictor for postoperative outcome [1] and misplacement rates up to 30% with non-navigated screw insertion have been described [2]. A range of navigation systems was made available, allowing image-guided screw insertion, intended to reduce the risk of pedicle perforation [3] and increase accuracy in placement of lumbar PS [4, 5]. However, drawbacks of these techniques have been considered: investment costs, lack of availability and training opportunities, and increased time-consumption due to interruption of the work flow have been pointed out [6]. There are also prospective studies and meta-analysis, which failed to show significant improvement of accuracy or clinical advantage in reduction of neurological symptoms of computer-navigated PS placement techniques [7‐9]. Still computer-navigated PS placement seems recommendable in especially complicated spine surgeries, due to significant deformity, revision surgeries, or complex trauma, to ensure accuracy [3, 10, 11].

Published learning curve analysis revealed, that computer-navigated PS placement offers high accuracy of lumbar PS placement from the beginning [12] and radiation exposure to the surgeon is reduced [13]. Based on these findings computer navigation may be favourable for novice operators to obtain initial experience in spine surgery, offering real-time control and allowing broad adjustment of arrangement during PS placement without increasing radiation exposure. However, the exact feasibility of computer navigation in training concerning safety, accuracy, and development of judgmental skills is still elusive.

The aim of our study was therefore to evaluate the feasibility of computer navigation as training method for PS placement, concerning safety, accuracy, and development of judgmental skills. The hypothesis was that the observed parameters would significantly improve for the trainee using navigation for PS placement.

A total of 138 screws were placed in 53 lumbar and 16 sacral segments of 18 patients (age 67.1 ± 15.1 years; range 42–85 years). No intra- or postoperative complications or instance of postoperative neurologic deficit or vascular injury were observed within a follow-up of at least 4 months. One patient had to be revised for implant dislocation mainly caused by poor bone quality which occurred 4 months after operation. A significant improvement of screw placement accuracy score according to Sclafani et al. [14] could be explored for the trainee between the initial and the final nine patients (8.2, 8.83; p = 0.006; Table 1; Fig. 1). Although, by trend, the trainee improved cortical pedicle violation rate, reaching similar values to the senior surgeon during the study, no significant differences were found in the percentage of PS totally contained within the pedicle according to Gertzbein et al. [2] and screw placement accuracy score between the two surgeons in the two subgroups (Table 1). A significant improvement of PS insertion time was explored for the trainee for lumbar PS placement between the initial and the final half of patients (13.3 min; 6.8 min; p = 0.005), whilst the placement time for the senior surgeon remained on a steady state (5.2 min; 4.1 min; p = 0.39) (Table 1; Fig. 2). Still, a significant difference in PS placement time between the two surgeons in initial (p = 0.001) and final subgroup (p = 0.049) remained (Table 1). A similar effect could be explored for sacral screws, even though the trainee’s improvement in placement time was not significant (Table 1). Although time per screw curves between surgeons seemed to exhibit a similar course (Fig. 2), no significant correlation was found between the two surgeons screw placement curves (r = 0.17; p = 0.5). Rate of PS placed by the trainee, that needed to be revised after initial placement, improved from 5.13 to 3.33% between initial and final nine patients (Table 1). Freehand PS placement in solid foam lumbar model at the end of the study significantly improved concerning cortical pedicle violation rate (1, 0.3; p = 0.001) screw placement accuracy score (7.0, 8.2; p = 0.003) with similar placement time compared to the performance at the beginning of the study (Table 2; Fig. 3).

After initial euphoria regarding computer navigation techniques for PS placement, chastening results concerning superiority of navigated techniques, regarding accuracy of pedicle screw placement, clinical outcome and economic aspects, have been explored and published by several authors [7, 9, 15]. The earlier explored low correlation between accuracy of screw placement, which will most evidently be improved with navigation, and clinical outcome in PS placement [16] might provide an explanation for the difficulty to effectively show a possible superiority. Some authors proposed that patient selection might play an important role for advantageous utilization of navigated PS insertion. Due to better overview on the current anatomy, navigation systems may be particularly required in complex cases as deformity, trauma, and revision surgeries [3, 10, 11].

In this study, we assumed that it might also depend on the type of surgeon, who is ‘selected’ to realize highest benefit from computer-navigated PS insertion. This assumption was based on earlier findings, that experienced surgeons achieve high PS accuracy using non-navigated PS insertion technique (thoracic spine; no pedicle breach in 97.5%) [17], whilst on the other hand for novice spine surgeons a learning interval with a clearly higher amount of inadequate PS placement has been described (cadaver, thoracic spine; adequate PS in 71.0%) [18]. This could vice versa implicate, that the highest benefit can be achieved by an unexperienced surgeon.

Therefore, an unexperienced surgeon was selected for PS placement using O-arm computer-navigated technology, to measure his improvement in PS placement accuracy and time. Permanent real-time control on the trainee during the screw positioning process provided by O-arm technology revealed in a low revision rate for failed screw placement, decreasing from an initially low level of 5.1% of PS to a revision rate of 3.3% at the end of the study. These results suggest that navigated PS insertion already provides a high safety for the patient from the early beginning of spine surgery training.

A highly adequate PS insertion result was reached by the trainee from the very first patient, with a steady increase, in the end appearing tangentially better, result compared to the experienced surgeon (Fig. 1). In this clinical setting, the trainee’s PS placement adequacy from the very beginning was comparable to earlier published reports on PS adequacy in patients in the literature [3, 14]. Significant improvement was also explored concerning enhancement in PS insertion time as shown by our data (Fig. 2).

Although this study did not focus on radiation dose to surgeon or patient, a reduction seems plausible when O-arm computer navigation technology is used for spine surgery training. We could show that adjustments and replacements, which—according to our experience—are performed by unexperienced surgeons to a higher extent, can be markedly reduced and withal, if necessary, do not cause additional radiation when O-arm technology is used.

Experience is based on judgmental skills according instrument handling, entry point direction, changes in resistance during hammering, and drilling as earlier experiments attempted to elaborate [18]. Still, it could not be closed out that permanent reliance on visual feedback provided by the computer navigation system counteracts the gain in judgmental skills in PS placement. Especially freehand placement of PS requires substantial judgmental skills and good anatomic knowledge [18]. Using the lumbar model at the beginning and end of the study for freehand PS placement by the trainee, mainly concerning on anatomical landmarks and tactile feedback during screw insertion, was our attempt to evaluate whether judgmental skills are gained using navigation technique, and whether they can be transferred to conventional screw placement. Our results from these experiments suggest that a substantial gain in judgmental skills could be elaborated with this technique, resulting in significantly improved PS positioning (Table 2; Fig. 3).

The use of O-arm technology in spine surgery training seems to provide a rapid gain in judgmental skills concerning PS placement, which might also be transferred to freehand PS placement. Adequacy of PS placement by a trainee using navigation was similar to screws placed by an experienced surgeon. The supervisor keeps permanent, real-time visual control on the PS placement process, offering a well-balanced combination of surgical training and patient safety. As conclusion we recommend that navigation technology as supportive technique should be used in spine surgery training whenever it is available, taking into consideration that exact anatomic skills are still required.