Multidimensional analysis of the learning curve for laparoscopic rectal cancer surgery

Background: The need for an initial learning experience in laparoscopic colorectal cancer surgery has been well established. However, the inherent differences in the complexity and results of laparoscopic rectal cancer surgery, as compared to colon surgery, warrant a study to analyze the learning curve exclusively for rectal cancer resections. Materials and Methods: four hundred thirty-one patients operated on between April 1994 and March 2006 were analyzed retrospectively for changes in surgical outcomes according to case sequence. A multidimensional analysis was done, based on the following parameters: conversion to laparotomy, intraoperative complications, postoperative complications, reoperations, operative time, and transfusion volumes. Multiple statistical methods were used for evaluation of the learning curve, which included the cumulative sum (CUSUM) method, risk-adjusted CUSUM, moving average method, and analysis of variance (ANOVA). Results: The risk factors for conversion were prior abdominal surgery (hazard ratio, 2.52; 95% CI, 1.04-6.10; P=0.04) and tumor size ≥3.5cm (hazard ratio, 5.05; 95% CI, 1.95-13.08; P=0.001). Risk-adjusted CUSUM analysis showed that case 61 was the peak change point for conversion. Postoperative complications occurred in 56 patients (13.0%), and the rate was associated significantly with case sequence (P<0.001). The turning point in the CUSUM model occurred at case 79, and the complication rates decreased thereafter. Operative time and intraoperative transfusion volumes stabilized over cases 61-75 and declined thereafter. Conclusions: Multidimensional analysis considering various surgical outcomes is necessary to evaluate the learning curve for laparoscopic rectal cancer surgery. The effective surgical learning curve was approximately 60-80 procedures in this series.