LAUSR

We are writing in response to the letter by Cardinale and Rizzi [1] regarding our recent manuscript in the Journal of Robotic Surgery, ‘‘The comparative accuracy of the ROSA stereotactic robot across a wide range of clinical applications and registration techniques’’ [2]. Unfortunately, there are several misunderstandings about our paper in this letter that should be addressed. First, the authors state that the accuracy in our paper is unacceptably low and related to clinical risk. While our paper was a focused, comparative paper, rather than a simple clinical case series, we can report that there were 0 complications or lead revisions or replacements due to inaccuracy in our patient cohort, leading to a clinical accuracy of 100%. Further, as we pointed out in our article, the major source of error in our cohort was placing the electrode/laser with a ‘deep’ bias in relationship to the planned target. This was done expressly to achieve clinical efficacy, that is, to ensure excellent laser fiber/electrode coverage of the target area. The fact that stereotactic accuracy can be lowered by a technique that improves clinical efficacy is perhaps one of the major messages of our report. Finally, in Table 7 of his own stereo-EEG accuracy paper, which he sites in his letter [3], Dr. Cardinale reports a range of target errors from 2.04 to 5.9 mm, putting the reported error of all of our applications squarely in the middle of the reported range of in vivo accuracy across a wide array of stereotactic devices including ROSA itself. To expand upon this further, to focus on the mean target error itself, rather the comparative accuracy across registration methods and applications is to miss the focus of our study. As we stated in the introduction of our report, our purpose was to compare techniques, not simply report our case series. Finally, Doctors Cardinale and Rizzi discuss the small sample size in our patient cohort and hypothesize that it is solely the reason for the lack of statistical difference between our different patient arms. This issue was acknowledged and reviewed by us in our discussion, but we will review it once more since it has led to some misunderstandings. Based on the difference of 0.6 mm (which was the largest mean difference in target error we reported) and using the standard deviation from our population, a power analysis showed that it would have taken as long as 10 years to collect sufficient data to show a statistically significant difference. This hardly seems worth it to detect a difference of less than 1 mm. This is to say nothing of the clinical significance. Even if it were true that bone fiducials are 0.6 mm more accurate, is that worth an extra procedure, head CT, and patient discomfort bone fiducials entail? As we reported, the likelihood that increasing the sample size would have changed the clinical result of that paper is extremely low, as we have shown, and so we feel the results as we reported have value to practicing surgeons. In conclusion, we would like to reiterate the fact that stereotactic accuracy is only one facet of the stereotactic procedure. Once clinical accuracy has been achieved at a high enough rate for a given clinical application, other & Nicholas J. Brandmeir [email protected]