Learning curve of robotic assisted total knee arthroplasty within a surgical team: A prospective study of 115 cases

Background

Over the last decade, robotic assistance has been highlighted to improve the accuracy of TKA

alignment. Despite growing adoption of robotic-assisted total knee arthroplasty (raTKA), little is known about the learning curve required to achieve consistent outcomes with personalized alignment techniques. The aim of this research was to assess the learning curve of robotic-assisted total knee arthroplasty (TKA) using the anatomic-functional implant positioning (AFIP) method by examining factors such as operative time, alignment accuracy, ability to restore the Coronal Plane Alignment of the Knee (CPAK) phenotype, postoperative ligament balance, and complication rate. This study addressed the following four specific research questions: (1) How many cases are necessary to improve operative time? (2) Does surgical experience affect implant positioning accuracy? (3) Does experience influence ligament balance and CPAK phenotype restoration? (4) Does increased experience improve patient-reported outcomes?

Hypothesis

Our hypothesis was that with increased experience using raTKA, operative time would improve, while maintaining high accuracy of implant positioning. We further hypothesized that personalized alignment targets, as defined by the AFIP protocol, including joint line obliquity and CPAK phenotype restoration, could be consistently achieved from the early cases, without being compromised during the learning phase.

Patients and methods

In total, 115 patients undergoing primary TKA with the ROSA robotic tool were prospectively included between February 2023 and February 2024. The AFIP technique was planned for each patient, and surgeries were performed by 4 experienced knee arthroplasty surgeons but with no prior experience in robotics. The following data were analyzed: (1) Preoperative and postoperative CPAK classification on full weight-bearing views. (2) Pre- and postoperative ligament balance, including the evaluation of medial and lateral femoro-tibial gaps in both flexion and extension. (3) The precision of postoperative frontal alignment (ΔHip Knee Angle (HKA)) was determined by calculating the discrepancy between the intended preoperative correction and the achieved postoperative adjustment. In addition complications and patient‐reported outcome measures (PROMs) were recorded for each patient. Cumulative summation (CUSUM) analysis was employed to evaluate the progression of learning curves.

Results

The implementation of raTKA required a learning period of 11 cases to achieve optimal operative time, with an average duration of 108 min.

Postoperative alignment remained stable throughout the series (ΔHKA = 2.0 ° ± 1.0 °), with no impact from surgical experience. Balanced medial-lateral gaps (defined as <2 mm difference) were achieved in 105 of 115 cases (91%). The CPAK phenotype was restored in 51 of 115 patients (44%), and JLO was successfully restored in 105 of 115 cases (91%). The rate of perioperative complications remained constant (3.5%, n = 4) and was not associated with case sequence (p = 0.89). PROMs improved significantly at 12 months (mean KSS from 57.1 ± 8.3 to 84.9 ± 9.2 [p < 0.001]).

Conclusion

Robotic-assisted TKA using the AFIP technique requires a short learning curve to optimize workflow, as operative time improves after 11 cases. Importantly, this learning period does not compromise implant positioning accuracy, CPAK and JLO restoration, ligament balance, or patient safety.

Level of evidence

IV; prospective investigation without control group.