Databases, Study Groups, and Evidence in Robotic Spine Surgery

Introduction

A large amount of literature related to spine surgery is plagued by retrospective design and small sample sizes resulting in a lack of statistical power. This is particularly true for literature regarding the validation of cutting-edge technologies, including robotic guidance systems. The methods by which researchers have attempted to provide studies of a higher yield than small institutional case series and cohort studies include the utilization of large administrative databases and forming multicenter study groups. This chapter will discuss the current evidence for robotic-assisted spine surgery and the importance of database utilization in research, particularly in the context of multicenter spinal robotics study groups.

Current Evidence for Robotic-Assisted Spine Surgery

The majority of literature available for robotic-assisted spine surgery centers on the accuracy of pedicle screw placement and is often compared to other techniques including free-hand and CT navigation. The Gertzbein and Robbins (GR) classification is typically used to grade the accuracy of screw placement, with accurate screws defined as those meeting either GR A (no breach) or B (breach <2 mm) criteria. The Ravi classification is also used. Numerous studies have reaffirmed the high accuracy of robotic-assisted pedicle screw placement since the FDA approval of the first robotic system in 2004. Multiple studies have verified the high accuracy of SpineAssist® (Mazor Robotics Ltd., Caesarea, Israel), with screws graded GR A or B ranging from 93.4% to 98.3%. A 2017 randomized controlled trial compared minimally invasive robotic-assisted and fluoroscopic-guided spinal fusions, utilizing the Renaissance® (Mazor Robotics Ltd., Caesarea, Israel) robotic guidance system. The study graded 97.7% of pedicle screws placed as GR A and 2.3% as GR B. Four studies have directly evaluated the accuracy of the Mazor X™ robotic guidance system (Medtronic, Dublin, Ireland) with clinical accuracy ranging from 97.5% to 100%. Further studies of recent generation robotic guidance systems have reaffirmed these high accuracy rates. At present, only seven studies have evaluated the accuracy of thoracolumbar pedicle screw placement with the current generation of navigated robotic guidance systems including the Mazor X Stealth Edition™ (XSE) (Medtronic, Dublin, Ireland) and ExcelsiusGPS™ (Globus Medical Inc, Audubon, Pennsylvania) ( Table 16.1 ). These studies report accuracy ranging from 97.3% to 100%.

Table 16.1

Summary of Accuracy Data Related to Current Generation Navigated Robotic Guidance Systems

Reference	Study Type	Device	Technique	Number of Patients	Number of Robotically Placed Screws	Screw Accuracy	Screw Grading
O’Connor et al.	Technical note	Mazor XSE	Not Specified	Not specified	90	100% (A)	GR
Godzik et al.	Retrospective case series	ExcelsiusGPS	Percutaneous	31	116	96.6% (1) / 2.59% (2)	Ravi
Jiang et al.	Retrospective cohort study	ExcelsiusGPS	Both – open and percutaneous	24	113	86.7% (A) /10.6% (B)	GR
Vardiman et al.	Retrospective case series	ExcelsiusGPS	Percutaneous	56	348	97.7% (A or B)	GR
Jain et al.	Retrospective case series	ExcelsiusGPS	Both – open and percutaneous	101	636	99% (A or B)	GR
Benech et al.	Retrospective case series	ExcelsiusGPS	Percutaneous	53	292	98.3% (A or B)	GR
Fayed et al.	Retrospective cohort study	ExcelsiusGPS	Percutaneous	20	100	94.2% (A) /3.88% (B)	GR

GR, Gertzbein and Robbins Classification.

Multiple studies exist comparing robotic-assisted surgery with CT navigation and fluoroscopic-guided techniques. These individual studies suggest equal to superior accuracy with robotic assistance, while minimizing radiation burden and complications. However, these studies are limited primarily by their retrospective nature and small sample sizes. This is confirmed by four systematic reviews with the consensus being that while robotic-assisted spine surgery may have potential benefits, the available literature contains significant limitations and heterogeneity that precludes a complete comparison with conventional techniques ( Table 16.2 ). Moreover, compared to determinations of accuracy, few studies have explored the postoperative complication rates and associated readmissions and surgical revisions. In fact, a recent meta-analysis comparing robotic assistance with CT navigation and fluoroscopic guidance for thoracolumbar instrumented fusion surgeries was unable to make any conclusions about differences in complication rates due to the lack of robotic data. Likely owing to a lack of single institutional data, some have turned to large administrative databases to explore this subject. The turn to database research has not been without challenges, however. For example, studies attempting to identify robotic-assisted procedures using billing codes will not be able to differentiate between robotic guidance and other forms of navigation, as a specific CPT code for robotic guidance does not currently exist. For this reason, and to be discussed in further detail later, an answer may lie in large multicenter institutional databases built with a focus on robotic surgery. Through our own inter-institution collaboration, we recently published the results of a retrospective analysis of a prospective database comparing free-hand with robotic-assisted thoracolumbar fusion surgery. We found an 83.20% reduction in the overall 90-day complication rate associated with robotic guidance ( P < .001). While these results show promise, further research into the overall postoperative complication, revision, and readmission rates, as well as long-term patient outcomes for robotic-assisted spine surgery will be necessary and likely take center stage in the near future.

Table 16.2

Summary of Systematic Reviews Evaluating Robotic Guidance Systems

Reference	Comparison Groups	Number of Studies	Findings	Conclusions
Yu et al.	RA, Free-hand	9	No differences in accuracy, complication rate, radiation exposure	Future well-designed studies are necessary to adequately compare techniques
Staartjes et al.	RA, Navigation, Free-hand	37	Decreased rate of postoperative screw revisions in navigation and RA compared to free-hand	RA has potential to reduce incidence of revisions related to screw malposition
Siccoli et al.	RA, Navigation, Free-hand	32	No differences in radiation, length of surgery, EBL, LOS, or complication rate for RA compared to free-hand	Findings attributable to a lack of statistical power; future high-quality studies required
Joseph et al.	RA	25	RA accuracy 85%–100%	RA associated with high degree of implant placement accuracy