The learning curve of robotically-assisted unicondylar knee arthroplasty

Summary Statement: Uptake of robotically-assisted orthopaedic surgery may be limited by a perceived steep learning curve. We quantified the technological learning curve and 5 surgeries were found to bring operating times to appropriate levels. Implant positioning was as planned from the outset. Introduction: Compared to total knee replacement, unicondylar knee replacement (UKR) has been found to reduce recovery time as well as increase patient satisfaction and improve range of motion. However, contradictory evidence together with revision rates concern may have limited the adoption of UKR surgery. Semi-active robotically-assisted orthopaedic tools have been developed to increase the accuracy of implant position and subsequent mechanical femorotibial angle to reduce revision rates. However, the perceived learning curve associated with such systems may cause apprehension among orthopaedic surgeons and reduce the uptake of such technology. To inform this debate, we aimed to quantify the learning curve associated with the technological aspects of the NavioPFS™ (Blue Belt Technologies Inc., Pittsburgh, USA) with regards to both operation time and implant accuracy. Methods: Five junior orthopaedic trainees volunteered for the study following ethical permission. All trainees attended the same initial training session and subsequently each trainee performed 5 UKR surgeries on left-sided synthetic femurs and tibiae (model 1146–2, Sawbones-Pacific Research Laboratories Inc, Vashon, WA, USA). A few days lapsed between surgeries, which were all completed in a two week window. Replica Tornier HLS Uni Evolution femoral and tibial implants (Tornier, France) were implanted without cementation. Each surgery was videoed and timings taken for key operation phases, as well as the overall operative time. A ball point probe with four reflective spherical markers attached was used to record the position of manufactured divots on the implant, which allowed the 3D position of the implant to be compared to the planned position. Absolute translational and rotational deviations from the planned position were analysed. Results: Total surgical time decreased significantly with surgery number (p < 0.001) from an initial average of 85 minutes to 48 minutes after 5 surgeries. All stages, except the cutting tool set up, demonstrated a significant difference in operative time with increasing number of surgeries performed (all p < 0.05) with the cutting phase decreasing from 41 to 23 minutes (p < 0.001). The translational and rotational accuracy of the implants did not significantly vary with surgery number. Discussion and Conclusion: The accuracy in implant position obtained by trainee surgeons on synthetic bones were similar to published data for experienced orthopaedic surgeons on other systems on cadavers. Whilst cadaver operations increase the complexity of operation, this should not theoretically affect the robotic system in preventing innaccurate implantation. Moreover, the fact that this accuracy was obtainable on the first surgery clearly demonstrates the system's ability in ensuring accurate implantation. Five surgeries dramatically reduced the total operative time, and moreover, the trend suggests that more surgeries would further decrease the total operation time. It was not the intention of the study to compare absolute trainee times on synthetic bones to surgeons with cadavers, but the learning curve of the protocol and technology suggests a halving of the operation time after 5 sessions would not be unrealisti