Agreement and Systematic Differences Between Intraoperative ROSA Hip-Knee-Ankle Measurements and Standing...

Background/Objectives: Intraoperative robotic hip-knee-ankle angle (HKA) measurements are obtained with the patient supine under operative conditions and may not directly represent postoperative standing weight-bearing alignment. This study quantified the correspondence and systematic differences between intraoperative ROSA HKA and standing radiographic HKA after robotic-assisted total knee arthroplasty. Methods: This retrospective single-center study included 208 knees from 178 consecutive patients, including 30 bilateral patients. The primary comparison was post-trial-component intraoperative ROSA HKA versus postoperative standing radiographic HKA; the secondary comparison was pre-cut intraoperative ROSA HKA versus preoperative standing radiographic HKA. Patient-cluster bootstrap confidence intervals used 10,000 resamples. Agreement was evaluated using Bland-Altman analysis, absolute-agreement ICC(A,1), and weighted kappa; Spearman ρ described association. Results: For the primary comparison, mean bias (ROSA minus radiographic HKA) was -1.75° (95% CI -2.08 to -1.43°), conventional limits of agreement were -6.36° to 2.85°, and ICC(A,1) was 0.472 (95% CI 0.379 to 0.557). Differences were within 3° in 70.2% of knees and exceeded 3° in 29.8%; proportional bias was present (slope -0.569, p < 0.001). The secondary comparison showed bias 0.97° and ICC(A,1) 0.956. Exact femoral and tibial size agreement was 42.3% and 80.8%, respectively. The standardized surgeon-entered 10 mm starting value was retained in 57.7% of knees. These systematic and angle-dependent differences may partly reflect the differing loading, positioning, and measurement conditions between the intraoperative trial construct and postoperative standing radiography. Conclusions: Post-trial-component ROSA HKA showed systematic and angle-dependent differences from postoperative standing HKA and should not be considered interchangeable with, or a fixed patient-level predictor of, postoperative standing alignment.