Surgeon Experience Influences Robotics Learning Curve for Minimally Invasive Lumbar Fusion: A Cumulative Sum Analysis.

Overview

abstract

STUDY DESIGN: Retrospective review of prospectively collected data. OBJECTIVE: To analyze the learning curves of three spine surgeons for robotic minimally invasive transforaminal lumbar interbody fusion (MI-TLIF). SUMMARY OF BACKGROUND DATA: Although the learning curve for robotic MI-TLIF has been described, the current evidence is of low quality with most studies being single-surgeon series. METHODS: Patients who underwent single-level MI-TLIF with three spine surgeons (years in practice: surgeon 1 - 4; surgeon 2 - 16; surgeon 3 - 2) using a floor-mounted robot were included. Outcome measures were operative time, fluoroscopy time, intraoperative complications, screw revision, and patient reported outcome measures (PROMs). Each surgeon's cases were divided into successive groups of 10 patients and compared for differences. Linear regression and cumulative sum (CuSum) analyses were performed to analyze the trend and learning curve, respectively. RESULTS: 187 patients were included (surgeon 1: 45, surgeon 2: 122, surgeon 3: 20). For surgeon 1, CuSum analysis showed a learning curve of 21 cases with attainment of mastery at case 31. Linear regression plots showed negative slopes for operative and fluoroscopy time. Both learning phase and post-learning phase groups showed significant improvement in PROMs. For surgeon 2, CuSum analysis demonstrated no discernible learning curve. There was no significant difference between successive patient groups in either operative time or fluoroscopy time. For surgeon 3, CuSum analysis demonstrated no discernible learning curve. Even though difference between successive patient groups was not significant, cases 11-20 had an average operative time 26 minutes less than cases 1-10), suggesting an ongoing learning curve. CONCLUSION: Surgeons who are well-experienced can be expected to have no or minimal learning curve for robotic MI-TLIF. Early attendings are likely to have a learning curve of around 21 cases with attainment of mastery at case 31. Learning curve does not seem to impact clinical outcomes following surgery. LEVEL OF EVIDENCE: 3.