Learning curves of robot-assisted pedicle screw fixations based on the cumulative sum test

Learning curve analyses in spine surgery: a systematic simulation-based critique of methodologies

BACKGROUND CONTEXT

Various statistical approaches exist to delineate learning curves in spine surgery. Techniques range from dividing cases into intervals for metric comparison, to employing regression and cumulative summation (CUSUM) analyses. However, their inherent inconsistencies and methodological flaws limit their comparability and reliability.

PURPOSE

To critically evaluate the methodologies used in existing literature for studying learning curves in spine surgery and to provide recommendations for future research.

STUDY DESIGN

Systematic literature review.

METHODS

A comprehensive literature search was conducted using PubMed, Embase, and Scopus databases, covering articles from January 2010 to September 2023. For inclusion, articles had to evaluate the change in a metric of performance during human spine surgery across time/a case series. Results had to be reported in sufficient detail to allow for evaluation of individual performance rather than group/institutional performance. Articles were excluded if they included cadaveric/nonhuman subjects, aggregated performance data or no way to infer change across a number of cases. Risk of bias was assessed using the Risk of Bias in Nonrandomized Studies of Interventions (ROBINS-I) tool. Surgical data were simulated using Python 3 and then examined via multiple commonly used analytic approaches including division into consecutive intervals, regression and CUSUM techniques. Results were qualitatively assessed to determine the effectiveness and limitations of each approach in depicting a learning curve.

RESULTS

About 113 studies met inclusion criteria. The majority of the studies were retrospective and evaluated a single-surgeon's experience. Methods varied considerably, with 66 studies using a single proficiency metric and 47 using more than 1. Operating time was the most commonly used metric. Interval division was the simplest and most commonly used method yet inherent limitations prevent collective synthesis. Regression may accurately describe the learning curve but in practice is hampered by sample size and model choice. CUSUM analyses are of widely varying quality with some being fundamentally flawed and widely misinterpreted however, others provide a reliable view of the learning process.

CONCLUSION

There is considerable variation in the quality of existing studies on learning curves in spine surgery. CUSUM analyses, when correctly applied, offer the most reliable estimates. To improve the validity and comparability of future studies, adherence to methodological guidelines is crucial. Multiple or composite performance metrics are necessary for a holistic understanding of the learning process.

Robot-assisted pedicle screw insertion versus navigation-based and freehand techniques for posterior spinal fusion in scoliosis: a systematic review and meta-analysis

PURPOSE

The role of robotics in spine surgery remains controversial, especially for scoliosis correction surgery. This study aims to assess the safety and efficacy of robotic-assisted (RA) surgery specifically for scoliosis surgery by comparing RA to both navigation systems (NS) and conventional freehand techniques (CF).

METHODS

As per the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines, a systematic review and meta-analysis were conducted via an electronic search of the following databases: MEDLINE, EMBASE, and the Cochrane Central Register of Controlled Trials (CENTRAL). All papers comparing RA to either NS or CF for posterior spinal fusion in scoliosis were included. Fixed and random effects models of analysis were utilised based on analysis heterogeneity.

RESULTS

10 observational studies were included in total. RA had significantly greater odds of accurate pedicle screw placement relative to both NS (OR = 2.02, CI = 1.52-2.67, p < 0.00001) and CF (OR = 3.06, CI = 1.79-5.23, p < 0.00001). The downside of RA was the significantly greater operation duration relative to NS (MD = 10.74, CI = 3.52-17.97, p = 0.004) and CF (MD = 40.27, CI = 20.90, p < 0.0001). Perioperative outcomes including estimated blood loss, radiation exposure, length of hospital stay, cobb angle correction rate, postoperative SRS score, VAS pain score, JOA score, as well as rates of neurological injury and revision surgery, were comparable between the groups (p > 0.05).

CONCLUSION

RA offers significantly greater pedicle screw placement accuracy relative to NS and CF, however, surgery can take longer. In terms of perioperative outcomes, all three techniques are comparable.

Learning Curve in Robotic Stereoelectroencephalography: Single Platform Experience

BACKGROUND

Learning curve, training, and cost impede widespread implementation of new technology. Neurosurgical robotic technology introduces challenges to visuospatial reasoning and requires the acquisition of new fine motor skills. Studies detailing operative workflow, learning curve, and patient outcomes are needed to describe the utility and cost-effectiveness of new robotic technology.

METHODS

A retrospective analysis was performed of pediatric patients who underwent robotic stereoelectroencephalography (sEEG) with the Medtronic Stealth Autoguide. Workflow, total operative time, and time per electrode were evaluated alongside target accuracy assessed via error measurements and root sum square. Patient demographics and clinical outcomes related to sEEG were also assessed.

RESULTS

Robot-assisted sEEG was performed in 12 pediatric patients. Comparison of cases over time demonstrated a mean operative time of 363.3 ± 109.5 minutes for the first 6 cases and 256.3 ± 59.1 minutes for the second 6 cases, with reduced operative time per electrode (P = 0.037). Mean entry point error, target point error, and depth point error were 1.82 ± 0.77 mm, 2.26 ± 0.71 mm, and 1.27 ± 0.53 mm, respectively, with mean root sum square of 3.23 ± 0.97 mm. Error measurements between magnetic resonance imaging and computed tomography angiography found computed tomography angiography to be more accurate with significant differences in mean entry point error (P = 0.043) and mean target point error (P = 0.035). The epileptogenic zone was identified in 11 patients, with therapeutic surgeries following in 9 patients, of whom 78% achieved an Engel class I.

CONCLUSIONS

This study demonstrated institutional workflow evolution and learning curve for the Autoguide in pediatric sEEG, resulting in reduced operative times and increased accuracy over a small number of cases. The platform may seamlessly and quickly be incorporated into clinical practice, and the provided workflow can facilitate a smooth transition.

Learning curves for robotic-assisted spine surgery: an analysis of the time taken for screw insertion, robot setting, registration, and fluoroscopy

PURPOSE

The purpose of this study was to clarify the learning curve for robotic-assisted spine surgery. We analyzed the workflow in robotic-assisted spine surgery and investigated how much experience is required to become proficient in robotic-assisted spine surgery.

METHODS

The data were obtained from consecutive 125 patients who underwent robotic-assisted screw placement soon after introducing a spine robotic system at a single center from April 2021 to January 2023. The 125 cases were divided into phases 1-5 of sequential groups of 25 cases each and compared for screw insertion time, robot setting time, registration time, and fluoroscopy time.

RESULTS

There were no significant differences in age, body mass index, intraoperative blood loss, number of fused segments, operation time, or operation time per segment between the 5 phases. There were significant differences in screw insertion time, robot setting time, registration time, and fluoroscopy time between the 5 phases. The screw insertion time, robot setting time, registration time, and fluoroscopy time in phase 1 were significantly longer than those in phases 2, 3, 4, and 5.

CONCLUSION

In an analysis of 125 cases after the introduction of the spine robotic system, the screw insertion time, robot setting time, registration time, and fluoroscopy time were significantly longer in the 25 cases in the period initially after introduction. The times were not significantly different in the subsequent 100 cases. Surgeons can be proficient in robotic-assisted spine surgery after their experience with 25 cases.

Learning curve of junior surgeons in robot-assisted pedicle screw placement: a comparative cohort study

OBJECTIVE

Robot-assisted technology has been gradually applied to pedicle screw placement in spinal surgery. This study was designed to detailedly evaluate the learning curve of junior surgeons in robot-assisted spine surgery.

METHODS

From December 2020 to February 2022, 199 patients requiring surgical treatment with posterior pedicle screw fixation were prospectively recruited into the study. The patients were randomized to the robot-assisted group (the RA group) or the conventional freehand group (the CF group). Under the senior specialist's supervision, pedicle screws were placed by two junior fellows without prior experience. Cumulative summation (CUSUM) analysis was performed on the learning curve of pedicle screw placement for performing quantitative assessment based on the time of screw insertion.

RESULTS

In total, 769 and 788 pedicle screws were placed in the RA and CF groups. Compared with the CF group, the learning duration in the RA group was shorter in the upper thoracic region (57 vs. 70 screws), but longer in the lower thoracic (62 vs. 58 screws) and the lumbosacral region (56 vs. 48 screws). The slope of learning curve was lower in the RA group than in the CF group. The screw accuracy in the RA group was superior to that in the CF group, especially in upper thoracic region (89.4% vs. 76.7%, P < 0.001). This disparity of accuracy became wider in deformity cases. In the upper thoracic region, the mean placement time was 5.34 ± 1.96 min in the RA group and 5.52 ± 2.43 min in the CF groups, while in the lower thoracic and lumbosacral regions, the CF group's mean placement times were statistically shorter. Three screw-related neural complications occurred in the CF group.

CONCLUSION

Robot-assisted technique has its advantages in the upper thoracic region and deformity cases, which is easier and safer to insert pedicle screws. The robot-assisted technique allowed a short learning curve for junior surgeons and exhibited consistently excellent results even in the early application period.

Learning Curves for Robot-Assisted Pedicle Screw Placement: Analysis of Operative Time for 234 Cases

BACKGROUND AND OBJECTIVES

Robot-assisted pedicle screw placement is associated with greater accuracy, reduced radiation, less blood loss, shorter hospital stays, and fewer complications than freehand screw placement. However, it can be associated with longer operative times and an extended training period. We report the initial experience of a surgeon using a robot system at an academic medical center.

METHODS

We retrospectively reviewed all patients undergoing robot-assisted pedicle screw placement at a single tertiary care institution by 1 surgeon from 10/2017 to 05/2022. Linear regression, analysis of variance, and cumulative sum analysis were used to evaluate operative time learning curves. Operative time subanalyses for surgery indication, number of levels, and experience level were performed.

RESULTS

In total, 234 cases were analyzed. A significant 0.19-minute decrease in operative time per case was observed (r = 0.14, P = .03). After 234 operations, this translates to a reduction in 44.5 minutes from the first to last case. A linear relationship was observed between case number and operative time in patients with spondylolisthesis (-0.63 minutes/case, r = 0.41, P < .001), 2-level involvement (-0.35 minutes/case, r = 0.19, P = .05), and 4-or-more-level involvement (-1.29 minutes/case, r = 0.24, P = .05). This resulted in reductions in operative time ranging from 39 minutes to 1.5 hours. Continued reductions in operative time were observed across the learning, experienced, and expert phases, which had mean operative times of 214, 197, and 146 minutes, respectively ( P < .001). General proficiency in robot-assisted surgery was observed after the 20th case. However, 67 cases were required to reach mastery, defined as the inflection point of the cumulative sum curve.

CONCLUSION

This study documents the long-term learning curve of a fellowship-trained spine neurosurgeon. Operative time significantly decreased with more experience. Although gaining comfort with robotic systems may be challenging or require additional training, it can benefit surgeons and patients alike with continued reductions in operative time.

Improving pedicle screw path planning by vertebral posture estimation

Objective.Robot-assisted pedicle screw placement in spinal surgery can reduce the complications associated with the screw placement and reduce the hospital return counts due to malfunctions. However, it requires accurate planning for a high-quality procedure. The state-of-the-art technologies reported in the literature either ignore the anatomical variations across vertebrae or require substantial human interactions. We present an improved approach that achieves pedicle screw path planning through multiple projections of a numerically re-oriented vertebra with the estimated posture.Approach.We proposed an improved YOLO-type neural network model (YOLOPOSE3D) to estimate the posture of a vertebra before pedicle path planning. In YOLOPOSE3D, the vertebral posture is given as a rotation quaternion and 3D location coordinates by optimizing the intersection over union of the vertebra with the predicted posture and the actual posture. Then, a new local coordinate system is established for the vertebra based on the estimated posture. Finally, the optimal pedicle screw path trajectory is determined from the multiple projections of the vertebra in the local coordinates.Main results.The experimental results in difficult cases of scoliosis showed that the new YOLOPOSE3D network could accurately detect the location and posture of the vertebra with average translation and orientation errors as small as 1.55 mm and 2.55°. The screw path planning achieved 83.1% success rate without breaking the pedicle cortex for the lumbar vertebral L1-L5, which is better than that of a doctor's manual planning, 82.4%. With the clinical class A requirement to allow less than 2 mm out of the pedicle cortex, the success rate achieved nearly 100%.Significance.The proposed YOLOPOSED3D method can accurately determine the vertebral postures. With the improved posture prior, better clinical outcomes can be achieved for pedicle screw placement in spine internal fixation procedures.

Learning curves in robotic neurosurgery: a systematic review

The transition to performing procedures robotically generally entails a period of adjustment known as a learning curve as the surgeon develops a familiarity with the technology. However, no study has comprehensively examined robotic learning curves across the field of neurosurgery. We conducted a systematic review to characterize the scope of literature on robotic learning curves in neurosurgery, assess operative parameters that may involve a learning curve, and delineate areas for future investigation. PubMed, Embase, and Scopus were searched. Following deduplication, articles were screened by title and abstract for relevance. Remaining articles were screened via full text for final inclusion. Bibliographic and learning curve data were extracted. Of 746 resultant articles, 32 articles describing 3074 patients were included, of which 23 (71.9%) examined spine, 4 (12.5%) pediatric, 4 (12.5%) functional, and 1 (3.1%) general neurosurgery. The parameters assessed for learning curves were heterogeneous. In total, 8 (57.1%) of 14 studies found reduced operative time with increased cases, while the remainder demonstrated no learning curve. Six (60.0%) of 10 studies reported reduced operative time per component with increased cases, while the remainder indicated no learning curve. Radiation time, radiation time per component, robot time, registration time, setup time, and radiation dose were assessed by ≤ 4 studies each, with 0-66.7% of studies demonstrated a learning curve. Four (44.4%) of 9 studies on accuracy showed improvement over time, while the others indicated no improvement over time. The number of cases required to reverse the learning curve ranged from 3 to 75. Learning curves are common in robotic neurosurgery. However, existing studies demonstrate high heterogeneity in assessed parameters and the number of cases that comprise the learning curve. Future studies should seek to develop strategies to reduce the number of cases required to reach the learning curve.

Results of single-incision distal biceps tendon repair for early-career upper-extremity surgeons

Background

The purpose of this investigation was to assess surgical outcomes after distal biceps tendon (DBT) repair for upper-extremity surgeons at the beginning of their careers, immediately following fellowship training. We aimed to determine if procedure times, complication rates, and clinical outcomes differed during the learning curve period for these early-career surgeons.

Methods

All cases of DBT repairs performed by 2 fellowship-trained surgeons from the start of their careers were included. Demographic data as well as operative times, complication rates, and patient reported outcomes were retrospectively collected. A cumulative sum chart (CUSUM) analysis was performed for the learning curve for both operative times and complication rate. This analysis continuously compares performance of an outcome to a predefined target level.

Results

A total of 78 DBT repairs performed by the two surgeons were included. In the CUSUM analysis of operative time for surgeon 1 and 2, both demonstrated a learning curve until case 4. In CUSUM analysis for complication rates, neither surgeon 1 nor surgeon 2 performed significantly worse than the target value and learning curve ranged from 14 to 21 cases. Mean Disabilities of Arm, Shoulder, and Hand score (QuickDASH) (10.65 ± 5.81) and the pain visual analog scale scores (1.13 ± 2.04) were comparable to previously reported literature.

Conclusions

These data suggest that a learning curve between 4 and 20 cases exists with respect to operative times and complication rates for DBT repairs for fellowship-trained upper-extremity surgeons at the start of clinical practice. Early-career surgeons appear to have acceptable clinical results and complications relative to previously published series irrespective of their learning stage.

The Learning Curve of Robotic-Assisted Pedicle Screw Placements Using the Cumulative Sum Analysis: A Study of the First 50 Cases at a Single Center

Introduction

The purpose of this study was to clarify how many cases surgeons need to experience to pass the learning phase of robotic-assisted spine surgery using the cumulative sum (CUSUM) analysis.

Methods

A retrospective review was conducted on the initial 50 consecutive patients who underwent robotic-assisted pedicle screw placements with open procedures using a spine robotic system (Mazor X Stealth Edition) at a single center from April 2021 to January 2022. There were 19 male and 31 female patients with a mean age of 58.7 (range, 13-86) years. To split the surgeries into the early and late phases using the CUSUM analysis of screw insertion time, we compared the screw insertion time, the robot setting time, the registration time, and the operation time in the early and late phases.

Results

The screw insertion time, the robot setting time, and the registration time declined as the number of surgical cases increased. The operation time did not decline as the number of surgical cases increased. The learning curve for screw insertion time can be separated into two stages based on the CUSUM analysis. The first 23 cases were in the early phase, and the later 27 cases were in the late phase. The mean screw insertion time was reduced from 3.2 min in the first 23 cases to 2.7 min in the subsequent 27 cases. The robot setting time and registration time in the late phase were also significantly shorter than those in the early phase.

Conclusions

The screw insertion time, robot setting time, and registration time decreased with experience. After 23 cases, surgeons passed the learning phase of robotic-assisted spine surgery and became more proficient.