Dear Editor, We commend Soni et al. for presenting one of the largest single-center prospective experiences evaluating robotic-assisted thoracolumbar pedicle screw placement, with a clearly described workflow and a high reported accuracy rate. Such real-world data are valuable as robotic systems become increasingly integrated into routine spine practice. Nevertheless, several aspects related to breach recognition, statistical methodology, radiation reporting, and outcome assessment warrant clarification to allow accurate interpretation of the findings. Screw accuracy appears to have been assessed using intraoperative three-dimensional O-arm imaging, with malpositioned screws replanned and revised during the same surgical setting. Although two independent radiologists are reported to have evaluated screw accuracy, it remains unclear whether they were involved during every surgery or reviewed images offline, whether they were blinded to surgical details, and whether all screws or only suspected breaches were assessed. Clarification of these methodological aspects would strengthen confidence in the reported accuracy rates. In addition, the manuscript reports mean radiation dose and O-arm usage time but does not explicitly state whether these values account for both the preinstrumentation scan used for planning and any subsequent confirmatory scans performed to assess screw position. If repeat intraoperative O-arm imaging was routinely performed, this would have implications for cumulative radiation exposure.1 Alternatively, if rescanning was selective based on intraoperative suspicion, the criteria triggering repeat imaging would be informative. Explicit reporting of the number and timing of O-arm scans per patient would enhance transparency and facilitate comparison with existing literature. Furthermore, while breaches are categorized by location and presumed cause, the manuscript does not indicate whether any breaches were associated with neurological symptoms or changes in neuromonitoring, information that would add important clinical context to the radiographic findings. From a statistical perspective, logistic regression was applied despite only 22 significant breach events (0.8%), a scenario in which multivariable modeling may be unstable.2 Although model fit was assessed using the Hosmer–Lemeshow test and pseudo-R² statistics, these measures are difficult to interpret in the setting of extreme class imbalance and when multiple screws are clustered within individual patients. Interrater reliability is reported as “Cohen’s kappa ≥0.8,” but the actual kappa value and confidence interval are not provided, limiting independent assessment of agreement robustness. Moreover, several factors explicitly identified by the authors as contributors to breach, such as mounting strategy, registration error, soft-tissue pressure, thoracic level anatomy, and learning-curve effects, were not incorporated into predictive modeling, despite apparent availability within the dataset. Finally, reporting of postoperative outcomes would benefit from greater temporal clarity. A mean postoperative Visual Analog Scale (VAS) score of 7.3 ± 2.5 is reported without specification of the time point of assessment, baseline pain levels, or pain subtype, and Table 3 further lists “VAS” under parameter and observation without clear differentiation. Addressing these points would refine and enhance transparency, and better align the study’s conclusions with its methodological framework. We offer these comments in a constructive spirit and believe that such clarifications would further strengthen an otherwise important contribution to the robotic spine surgery literature. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest.
Rangaswamy et al. (Sun,) studied this question.