Learning curve of two-incision thoracoscopic double valve replacement: Cumulative sum analysis of postoperative outcomes and proficiency in 252 operations from a single high-volume center
Qichen Xu, Zheng Xu, Liangwan Chen, Xiaofu Dai, Zhiqin LinBackground
Despite the increasing adoption of minimally invasive approaches for valvular heart disease, the learning trajectory for complex thoracoscopic double valve replacement (DVR) remains largely uncharacterized.
Objectives
To evaluate the learning curve of two-incision thoracoscopic DVR, determine the case threshold for procedural proficiency, and identify independent predictors of major adverse surgical events (MASE).
Design
Retrospective cohort study.
Methods
We evaluated 252 consecutive thoracoscopic DVR procedures performed at a single high-volume center between March 2017 and December 2024. Cumulative sum (CUSUM) analysis of cardiopulmonary bypass (CPB) time was utilized to define distinct learning phases. MASE, a composite of mortality, conversion to sternotomy, and reoperation for bleeding, served as the primary clinical outcome. Multivariate logistic regression was performed to identify independent predictors of MASE.
Results
CUSUM analysis revealed three distinct learning phases: Acquisition (cases 1–67), Acceleration (cases 68–139), and Mastery (cases 140–252). Mean CPB time decreased progressively from 187.5±34.2 to 156.3±27.6 and 138.4±18.9 minutes across the three phases (p<0.001). Concurrently, MASE rates declined significantly from 16.4% in the Acquisition phase to 4.2% in the Acceleration phase, stabilizing at 4.4% during the Mastery phase (p=0.006). A small aortic annulus emerged as the strongest independent predictor of complications (OR 467.483, p<0.001), followed by operations performed during the Acquisition phase (OR 14.564, p=0.007).
Conclusion
Thoracoscopic DVR demonstrates a three-phase learning curve, with technical proficiency achieved after approximately 139 cases. Early procedural implementation requires risk-adapted case selection, specifically avoiding small aortic annuli during the initial experience. Because standard risk models may significantly underestimate mortality in this context, procedure-specific considerations and learning curve position are essential when evaluating candidates for this advanced minimally invasive approach.