A novel robotic surgical assistant for total knee arthroplasty has a learning curve ranging from 6 to 14 cases and exhibits high accuracy in tibial bone cuts

Abstract

Abstract Background The adoption of robot-assisted total knee arthroplasty (TKA) aims to enhance the precision of implant positioning and limb alignment. Despite its benefits, the adoption of such technology is often accompanied by an initial learning curve, which may result in increased operative times. This study sought to determine the learning curve for the ROSA (Robotic Surgical Assistant) Knee System (Zimmer Biomet) in performing TKA and to evaluate the accuracy of the system in executing bone cuts and angles as planned. The hypothesis of this study was that cumulative experience with this robotic system would lead to reduced operative times. Additionally, the ROSA system demonstrated reliability in terms of the accuracy and reproducibility of bone cuts. Methods In this retrospective observational study, we examined 110 medical records from 95 patients who underwent ROSA-assisted TKA performed by three surgeons. We employed the cumulative summation methodology to assess the learning curves related to operative time. Furthermore, we evaluated the accuracy of the ROSA Knee System in performing TKA by comparing planned versus validated values for femoral and tibial bone cuts and angles. Results The learning curve for the ROSA Knee System spanned 14, 14, and 6 cases for the respective surgeons, with operative times decreasing by 22 min upon reaching proficiency (70.8 vs. 48.9 min; p < 0.001). Significant discrepancies were observed between the average planned and validated cuts and angles for femoral bone cuts (0.4 degree ± 2.4 for femoral flexion, 0.1 degree ± 0.6 for femoral coronal alignment, 0.3 mm ± 1.2 for distal medial femoral resection, 1.4 mm ± 8.8 for distal lateral femoral resection) and hip–knee–ankle axis alignment (0.3 degree ± 1.9 )(p < 0.05) but not for tibial bone cuts. Differences between planned and validated measurements during the learning and proficiency phases were nonsignificant across all parameters, except for the femoral flexion angle (0.42 degree ± 0.8 vs. 0.44 degree ± 2.7) (p = 0.49). Conclusion The ROSA Knee System can be integrated into surgical workflows after a modest learning curve of 6 to 14 cases. The system demonstrated high accuracy and reproducibility, particularly for tibial bone cuts. Acknowledging the learning curve associated with new robot-assisted TKA technologies is vital for their effective implementation.

Keywords

• Accuracy
• Cumulative summation analysis
• Learning curve
• Operative time
• Robot-assisted system
• ROSA knee system