Login to Access Video or Poster Abstract: MP49-20
Sources of Funding: none

Introduction

One of the presumed benefits of robot-assisted partial nephrectomy (RPN) would be a shorter learning curve compared to laparoscopic partial nephrectomy (LPN). To date, no study has aimed to assess the learning curve of RPN through a multivariate model, which has become the most accepted approach to analyze learning curve of surgical procedures The objective of this study was to evaluate the learning curve of the RPN through a multivariate model

Methods

The data of all patients who underwent an RPN in a single-center between 2010 and 2016 were collected prospectively. All RPN carried out by one of the surgeons were included in this study. The primary endpoint was the TRIFECTA defined as follows: no complications and warm ischemia time<25 minutes and negative surgical margins. According to the literature, the TRIFECTA achievement rate of an expert surgeon is 65% (Kalifeh et al, J Urol 2013). The achievement of TRIFECTA was analysed using a multivariate logistic regression model adjusting for surgeon experience, patients and tumors characteristics._x000D_

Results

Two-hundred and one consecutive RPN performed by a single-surgeon were included in the study. Positive surgical margins rate was 3.4%, complications rate was 25.9% and mean warm ischemia time was 15.4 minutes. The TRIFECTA achievment rate was 67.2% when considering the whole cohort. According to the multivariate model, the « expert » level was reached after 60 procedures (figure 1). However, when performing a bootstraping projection, the TRIFECTA achievement rate would have continued to improve beyond the 200th procedure and would stabilize only after 500 procedures

Conclusions

In this study, the first to appropriately analyze the RPN learning curve through multivariate regression model, 60 procedures were necessary to obtain the outcomes of an expert surgeon. The boostraping allowed to estimate that perioperative outcomes could keep improving until the 500th procedure suggesting a longer learning curve than presumed.

Funding

none