Surg Endosc. 2026 Jul 27. doi: 10.1007/s00464-026-13196-4. Online ahead of print.
ABSTRACT
BACKGROUND: Robotic cholecystectomy is increasingly adopted as an alternative to laparoscopic cholecystectomy and proposed as an entry‑level procedure in robotic training curricula, yet real‑world data on surgeon‑specific learning curves and their impact on perioperative safety remain limited. This study aimed to map the learning curve for robotic cholecystectomy across a multi-surgeon, multicenter cohort and assess associated safety outcomes.
METHODS: This retrospective cohort study analyzed the first 50 consecutive robotic cholecystectomies performed independently by each of five surgeons (n = 250) between January 2023 and June 2025. For each surgeon, the learning-curve endpoint was identified as the breakpoint of a piecewise linear regression of skin-to-skin operative time against case sequence, and the Mann-Whitney U test assessed whether operative time differed significantly between the early and late phases. A linear mixed-effects model, with surgeon as random intercept, identified independent predictors of operative time accounting for within-surgeon clustering.
RESULTS: The learning-curve endpoint ranged from 11 to 34 cases across surgeons. The breakpoint corresponded to a significant reduction in operative time for three of five surgeons (p = 0.017, p = 0.015, p < 0.001), but not for the remaining two (p = 0.853, p = 0.233). Cohort-level median operative time decreased significantly from early to late phase (66 vs 50 min; p < 0.001), while length of stay did not differ (p = 0.354). Case sequence number (- 0.72 min/case) and Nassar difficulty grade (+ 9.8 min/grade) were independent predictors of operative time (p < 0.001). No bile duct injuries occurred and severe complications occurred in 1/250 patients (0.4%).
CONCLUSIONS: Most surgeons showed a statistically confirmed reduction in operative time within their first 11-24 robotic cholecystectomies, while for others no significant improvement was confirmed. Complication rates remained low, although this cohort was not adequately powered to formally demonstrate safety equivalence across the learning process. These findings support robotic cholecystectomy as a feasible early procedure within structured robotic training pathways.
PMID:42509501 | DOI:10.1007/s00464-026-13196-4
