Mediolateral Joint Balance Assessment Is Reproducible Between Surgeons, Under No Standardized Varus–Valgus Stress, During Robot-Assisted Total Knee Arthroplasty

Abstract

Objective: To evaluate inter-surgeon variability in mediolateral joint balance assessment during robotic-assisted total knee arthroplasty (TKA) using the robotic surgical assistant (ROSA) Knee System, and to determine whether such variability influences the surgical planning generated by this system. Methods: An observational study was conducted including 101 patients undergoing primary robotic-assisted TKA. Two experienced surgeons independently assessed intraoperative mediolateral laxity by manually applying varus–valgus stress across six predefined flexion angles (0°, 30°, 45°, 60°, 90°, and 120°). Measurements of medial and lateral gaps (mm) and varus–valgus alignment (°) were recorded using the ROSA Knee System. Inter-observer differences were analysed using the Wilcoxon signed-rank test, and reliability was assessed using the intraclass correlation coefficient (ICC). The impact of inter-surgeon variability on surgical planning, including bone resection and compartmental space measurement parameters, was also evaluated. Results: No statistically significant inter-observer differences were observed in mediolateral gap measurements or varus–valgus alignment across all flexion angles (p > 0.05). Inter-observer reliability for varus–valgus measurements was good to excellent in extension and early flexion (ICC 0.86–0.91), decreased to moderate at 90° (ICC 0.68–0.74) and was poor for varus assessment at 120° (ICC 0.46, 95% CI 0.29–0.60). For gap measurements, reliability was excellent for the medial gap and good for the lateral gap in extension (ICC 0.93 and 0.89, respectively), and moderate for both gaps at 90° (ICC 0.71–0.74). No significant inter-observer differences were found in cut or space measurements, with excellent agreement for cut parameters (ICC > 0.95). Conclusions: Inter-surgeon assessment of mediolateral joint balance during robot-assisted TKA showed good-to-excellent reproducibility in extension and early flexion, with reduced reproducibility in deep flexion. Statistically significant inter-observer differences were not observed in the resulting cut and space measurements.