To characterize learning-curve phases in robot-assisted radical prostatectomy (RARP) through detailed temporal outcome analysis and to establish evidence-based proficiency milestones for surgical training programs during early independent practice after fellowship training. Retrospective analysis of 756 consecutive RARP cases performed by two fellowship-trained surgeons (2014–2020) at a single tertiary institution. Surgeon A (512 cases, 147 prior fellowship cases) and Surgeon B (244 cases, 89 fellowship cases) performed all procedures. Primary outcomes included operative time, estimated blood loss, positive surgical margins, and complications. Learning-curve assessment employed moving-average analysis (20-case window), cumulative sum (CUSUM) methodology, and both phase-based analysis (Phase 1: cases 1–75, n = 75; Phase 2: cases 76–200, n = 125; Phase 3: cases 201–756, n = 556) and quartile-based comparison for temporal outcome evolution. Restricted cubic spline regression was used to model nonlinear learning dynamics, and surgeon-level CUSUM analyses were performed to assess concordance between individual learning trajectories and pooled findings. Three distinct learning-curve phases emerged, with CUSUM inflection points at cases 82 and 178. Phase-based analysis demonstrated: Phase 1 (n = 75) mean operative time 218.4 ± 46.3 min, estimated blood loss 328.7 ± 184.2 mL, positive margins 16.0% (12/75), complications 6.7% (5/75); Phase 2 (n = 125) operative time 189.2 ± 39.8 min, blood loss 282.4 ± 162.1 mL, positive margins 12.0% (15/125), complications 4.0% (5/125); Phase 3 (n = 556) operative time 174.6 ± 38.2 min, blood loss 248.3 ± 148.9 mL, positive margins 9.0% (50/556), complications 1.8% (10/556). From Phase 1 to Phase 3, operative time decreased by 20.1% (p < 0.001), blood loss by 24.5% (p < 0.001), positive margins by 43.8% (p = 0.041), and complications by 73.1% (p = 0.024). Bilateral nerve-sparing increased across phases (41.3% to 56.8%; p = 0.022), and erectile function recovery improved (58.3% to 78.9%; p = 0.003), while urinary continence remained uniformly high without a significant phase effect (p = 0.089). Restricted cubic spline regression confirmed a nonlinear, asymptotic improvement pattern, with early rapid gains and a plateauing pattern consistent with Phase 3 performance stabilization. The RARP learning curve extends through approximately 200 cases and comprises three distinct phases that require differentiated training strategies. The duration of learning beyond typical fellowship case volumes supports the need for structured early-career mentorship and multidimensional outcome assessment, including functional endpoints and nerve-sparing practice patterns. This three-phase model provides an actionable framework for training program design, defining Phase 1 completion (case 75) as basic competency, Phase 2 completion (case 200) as advanced proficiency, and Phase 3 (cases > 200) as performance stabilization (mature proficiency plateau).

Evolution of robotic prostatectomy techniques: impact of nerve preservation and surgical experience on oncological and functional outcomes

期刊论文
发表日期：

2026年4月2日

查看全文

Walaa M. Borhan,
Emad Rajih,
Yasir Hassan Elhassan,
Assaad El-Hakim

摘要

To characterize learning-curve phases in robot-assisted radical prostatectomy (RARP) through detailed temporal outcome analysis and to establish evidence-based proficiency milestones for surgical training programs during early independent practice after fellowship training. Retrospective analysis of 756 consecutive RARP cases performed by two fellowship-trained surgeons (2014–2020) at a single tertiary institution. Surgeon A (512 cases, 147 prior fellowship cases) and Surgeon B (244 cases, 89 fellowship cases) performed all procedures. Primary outcomes included operative time, estimated blood loss, positive surgical margins, and complications. Learning-curve assessment employed moving-average analysis (20-case window), cumulative sum (CUSUM) methodology, and both phase-based analysis (Phase 1: cases 1–75, n = 75; Phase 2: cases 76–200, n = 125; Phase 3: cases 201–756, n = 556) and quartile-based comparison for temporal outcome evolution. Restricted cubic spline regression was used to model nonlinear learning dynamics, and surgeon-level CUSUM analyses were performed to assess concordance between individual learning trajectories and pooled findings. Three distinct learning-curve phases emerged, with CUSUM inflection points at cases 82 and 178. Phase-based analysis demonstrated: Phase 1 (n = 75) mean operative time 218.4 ± 46.3 min, estimated blood loss 328.7 ± 184.2 mL, positive margins 16.0% (12/75), complications 6.7% (5/75); Phase 2 (n = 125) operative time 189.2 ± 39.8 min, blood loss 282.4 ± 162.1 mL, positive margins 12.0% (15/125), complications 4.0% (5/125); Phase 3 (n = 556) operative time 174.6 ± 38.2 min, blood loss 248.3 ± 148.9 mL, positive margins 9.0% (50/556), complications 1.8% (10/556). From Phase 1 to Phase 3, operative time decreased by 20.1% (p < 0.001), blood loss by 24.5% (p < 0.001), positive margins by 43.8% (p = 0.041), and complications by 73.1% (p = 0.024). Bilateral nerve-sparing increased across phases (41.3% to 56.8%; p = 0.022), and erectile function recovery improved (58.3% to 78.9%; p = 0.003), while urinary continence remained uniformly high without a significant phase effect (p = 0.089). Restricted cubic spline regression confirmed a nonlinear, asymptotic improvement pattern, with early rapid gains and a plateauing pattern consistent with Phase 3 performance stabilization. The RARP learning curve extends through approximately 200 cases and comprises three distinct phases that require differentiated training strategies. The duration of learning beyond typical fellowship case volumes supports the need for structured early-career mentorship and multidimensional outcome assessment, including functional endpoints and nerve-sparing practice patterns. This three-phase model provides an actionable framework for training program design, defining Phase 1 completion (case 75) as basic competency, Phase 2 completion (case 200) as advanced proficiency, and Phase 3 (cases > 200) as performance stabilization (mature proficiency plateau).