The second randomized control trial was undertaken by Kim et al. [9] in patients undergoing monosegment posterior lumbar interbody fusion (PLIF), for degenerative or spondylotic spondylolisthesis and central canal stenosis, through RA or FH techniques. Accuracy was assessed using a cumulative summation test (CUSUM). The CUSUM test is a quality control monitor that has been implemented in a number of different surgical settings and is particularly useful in assessing surgical performance and learning curve during the implementation of a new procedure or technique [10]. In total 80 pedicle screws were inserted into 20 patients in each of the RA and FH arms of the study. Pedicle screw placement was assessed based on the Gertzbein-Robbins scale applied to post-operative CT scans. Results showed that there were no significant differences in baseline demographics between the cohorts pre-operatively. Screw placement entirely within the pedicle (grade A) was not statistically significant with placement in 95% of RA and 91% of FH pedicle screws. The main difference between this and the study by Ringel et al. is the consideration of only grade A pedicle screw placement as satisfactory and the addition of the Peteron technique. This involves the introduction of a specially designed Peteron instrument with teeth at the distal end. The instrument is used to flatten and smooth the proposed entry point of the screw such that drilling can proceed without lateral skidding. Furthermore, a more recent software upgrade allows visualization of the iliac crests during the planning stage such that the S1 pedicle screw trajectory is no longer impinged upon by the iliac crest and therefore not deflected medially. Kim et al. also used the ‘bed mount’ technique of robot stabilization in 85% of cases and the spine clamp in the remainder. Unlike Ringel et al. [8], they did not find that was a factor that contributed to inaccuracy. CUSUM analysis is also a means of assessing cumulative accuracy of surgical interventions when new technology is introduced to ensure that performance is no worse than the current standard of care. It has been employed as an early warning system in which when the cumulative performance crosses a particular decision threshold, the procedure is said to be ‘substandard’ and the procedure should be stopped or have corrective measures introduced to prevent subsequent patients from harm [10]. The use of such a statistical process control has been shown to detect inadequate performance much earlier than other statistical methods. The results of the CUSUM analysis in this randomized control trial revealed a steep learning curve that can be overcome, as in this case, through a brief manufacturer-sponsored training course and a training set of five clinical cases. Operative time, radiation exposure and post-operative complications and hospital stay were not assessed in this study.

An ongoing single-centre randomized control trial by Roser et al. [11] has to date only published interim results in 40 FH, 36 navigation-assisted (NA) and 72 RA pedicle screws in 37 patients. Accuracy of the pedicle screws was determined using the Gertzbein and Robbins scale. Only when the screw was entirely within the pedicle (position A) was this considered satisfactory. Ninety-eight percent of FH, 92% of navigation-assisted and 99% of RA pedicle screws were found to be satisfactory. The study aims to randomize a total of 120 patients (40 in each of the three arms). Radiation exposure was measured as both the total fluoroscopy time and the total radiation dose measured using a dosimeter. Radiation time was FH 31.5 s, NA 10.4 s and RA 16 s. Given that only one quarter of these have been recruited, the authors have not performed any statistical evaluation. The full results of the trial are therefore awaited.

Non-randomized cohort studies have been performed prospectively and retrospectively by Schizas et al. [12] and Kantelhardt et al. [13], respectively. Schizas et al. undertook a single-centre prospective cohort study in which 34 patients undergoing thoracolumbar pedicle screw fixation were divided into 64 screws in 23 patients in the FH group and 64 screws in 11 patients in the RA group. Two independent observers rated post-operative CT scans using the Rampersaud scale. Screw position either entirely within the pedicle or breaches within 3 mm of the pedicle wall (positions A and B) was considered satisfactory. This study found 92% of the FH and 95% of RA screws were satisfactory. This was not statistically significant. Radiation exposure was also not found to be significantly different. Kantelhardt et al. performed a retrospective cohort analysis comparing RA 250 pedicle screws in 35 percutaneous and 20 open cases with 286 FH pedicle screws in 57 open cases. In this study the Wiesner and Schizas scale was utilized in which screws that were entirely within the pedicle or with encroachment of the cortical bone (grade 0 and 1) were deemed satisfactory. This was achieved in 94.5% of RA and 91.5% of FH pedicle screws (p < 0.05). In total 1% of the RA compared to 12.2% of the FH pedicle screws required revision, although this was not statistically significant. Comparison of total radiation exposure, measured as total fluoroscopy time, was 27 s for percutaneous RA, 43 s for open RA and 77 s for open FH cases, indicating a significant reduction in surgeon and patient irradiation (p < 0.02). There was no different in the average time per screw placement between RA and FH cases in this study. Post-operatively this study revealed a significant increase in the requirement for opioid analgesics from 38% percutaneous RA and 67% open RA to 89% open FH (p < 0.004). Intraoperative dural tears were seen in 4.7% of RA and 9% of open FH cases whilst post-operative infections occurred in 2.7% RA and 10.7% open FH cases (p < 0.04). Total hospital stay was also reduced to 10.6 days in RA from 14.6 days in FH procedures (p = 0.009). Both of the non-randomized studies by Kantelhardt et al. and Schizas et al. failed to provide a prospective power calculation. In addition it is unclear whether there were any significant baseline differences in the patients in the RA and FH groups that could have impacted on the outcome.

Schatlo et al. [14] performed a retrospective matched cohort comparison of 244 RA pedicles screws placed in 55 patients and 163 FH pedicle screws in 40 patients. Of the patients in the RA group, they were subdivided into open RA in which 83 pedicle screws were inserted in 17 patients percutaneous RA in which 161 pedicle screws were inserted in 38 patients. The RA procedures employed the spinous process clamp method of attaching the robot to the patient. Baseline patient demographics were closely matched although there were fewer males in the open RA and BMI was higher in the open FH compared to the percutaneous RA group. Pedicle screw accuracy was assessed from post-operative CT scans by a blinded neuroradiologist using the Gertzbein and Robbins scales in which those that were entirely within the pedicle or with <2 mm breach (A and B) of the wall were “clinically acceptable”. There was no statistically significant difference between the RA and FH groups with 91.4% and 87.1% achieving satisfactory position. There was no difference in infection rate between the RA (1.8%) and the FH (2.5%) groups. In one case a FH screw resulted in a painful radiculopathy at the level of L4 and therefore required revision. Total operative time was non-significantly longer in the RA groups. Blood loss was significantly lower in the RA group falling from a mean of 375 ml to 713 ml (p < 0.01). There was no difference in the length of hospital stay between the two groups. Post-operative pain was assessed as the cumulative dose of morphine and did not show any significant difference. Similar to Ringel et al., Schatlo et al. describe a discrepancy in the planned and radiographic robot entry points on lateral fluoroscopy despite accurate registration and attribute this to the entry cannula sliding off the steep superior articular facet. They feel modification of the technique through choice of a flat entry point or tightly securing the cannula teeth on the bone may improve the lateral deviation errors. Overall radiation exposure was not recorded between the two techniques.