A systematic review of the learning curve in robotic surgery: range and heterogeneity

With the rapid adoption of the robotic surgery, more and more learning curve (LC) papers are being published but there is no set definition of what should constitute a rigorous analysis and represent a true LC. A systematic review of the robotic surgical literature was undertaken to determine the range and heterogeneity of parameters reported in studies assessing the LC in robotic surgery.

The search was conducted in July 2017 in PubMed. All studies reporting a LC in robotic surgery were included. 268 (25%) of the identified studies met the inclusion criteria.

102 (38%) studies did not define nor explicitly state the LC with appropriate evidence; 166 studies were considered for quantitative analysis. 46 different parameters of 6 different outcome domains were reported with a median of two parameters (1-8) and 1 domain (1-5) per study. Overall, three domains were only technical and three domains were both technical and clinical/patient-centered outcomes. The two most commonly reported domains were operative time [146 studies (88%)] and intraoperative outcomes [31 studies (19%)]. Postoperative outcomes [16 studies (9%)] and surgical success [11 studies (7%)] were reported infrequently. Purely technical outcomes were the most frequently used to assess LC [131 studies (79%)].

The outcomes reported in studies assessing LC in robotic surgery are extremely heterogeneous and are most often technical indicators of surgical performance rather than clinical and patient-centered outcomes. There is no single outcome that best represents the surgical success. A standardized multi-outcome approach to assessing LC is recommended.