Summary

A comparable surgical time to conventional surgery is reached after 21 robotic-assisted total knee arthroplasties.

Abstract

**Introduction:** One of the main challenges in the implementation of robotic assistance in total knee arthroplasty is the increase in surgical time (Tcx).
**Objective:** To determine whether it is possible to match the Tcx of robotic-assisted knee arthroplasty to that of the conventional technique, and at what point in the learning curve this is achieved.
**Method:** Cross-sectional study (longitudinal data). The Tcx from the first 50 robotic-assisted cases by a surgeon in a hospital setting were included. All revision cases and unicompartmental arthroplasties were excluded. Additionally, the Tcx from 75 consecutive primary conventional arthroplasties performed prior to the introduction of robotic assistance by the same surgeon were collected. The Tcx were obtained from the nursing records in the electronic medical chart and include the time from incision to dressing. A Markov switching model was estimated to establish the number of significant states in robotic surgery Tcx. A Wald test was then applied to identify the case where the shift occurred. Finally, the Tcx of robotic surgery and conventional surgery were compared using the Kruskal-Wallis test.
**Results:** Two significant states were identified, with a significant shift observed at case 21. The first state had a median Tcx of 118 minutes (range: 90 to 190), and the second, a median of 105 minutes (range: 65 to 180). Conventional surgery had a median Tcx of 100 minutes (range: 60 to 196). The robotic group in the first 20 cases was significantly different from the second group and from conventional surgery, with no differences found between the second robotic group and conventional surgery.
**Discussion/Conclusions:** Based on Tcx as a parameter, the estimated learning curve is 20 cases to reduce surgical times to levels comparable to conventional surgery.