BACKGROUND: To date, most studies describing the “learning curve” of Robotic Assisted Laparoscopic Prostatectomy (RALP) have used subjective measures of comfort and competence to state when the learning curve has been over-come. Notably lacking from these studies has been a uniform notion of what the learning curve is and what “overcoming it” means. We sought examples from industry to establish a more objective and useful methodology of assessing proficiency. Originally conceived for analysis of the aircraft industry during WWII, the “learning rate” represents the percentage decrease in production time per doubling of experience. When graphically represented, learning rates take the form of power functions and are called “learning curves”. Using this technique, we evaluated the average learning rate for RALP of three experienced open surgeons to establish when the learning rate had plateaued.
METHODS: Non-Randomized, prospective study. The first 75 RALPs performed by 3 surgeons from November 2003 until November 2006 were evaluated. Operative characteristics studied were Total Procedure Time (TPT), Estimated Blood Loss (EBL), Bladder Neck Contractures (BNC) and Positive Surgical Margins (PSM). Discrete variables were analyzed using the Fisher Exact Test; means of continuous variables were compared using two tailed t-tests. Learning curves were created using linear regression analysis, in the form of a power function, on scattergrams of the TPT for each of the three surgeons. From the equations of the best-fit lines, values for TPT were derived at consecutive doublings. Learning rates were found by determining the percentage change between these doublings of experience. Learning rates were averaged at the procedure intervals 1-75, 16-75, 21-75, 26-75 and 31-75.
RESULTS:
Table 1: Learning rates of the intervals evaluated

Table 2: Operative Characteristics

CONCLUSIONS: 1. We propose that improvements in efficiency are made at a baseline rate indefinitely. Accordingly, it is more meaningful to describe the point at which the learning rate has plateaued rather than when the learning curve has been overcome.
2. The average learning rate plateaus by the 25^th procedure and not by the 20^th procedure. It can thus be inferred that the learning rate has stabilized by the 25^th procedure but not the 20^th.
3. Given that there is no statistically significant difference in intra-operative characteristics between intervals 1-25 and 26-75, with the exception of TPT, analysis of surgical proficiency and competence is reasonably confined to procedure time.