Replication of the Robotic Evaluation of Articular Laxity (REAL) classification using intraoperative gap data in robot-assisted total knee arthroplasty.

Abstract

Robot-assisted total knee arthroplasty (TKA) enables objective assessment of mediolateral (ML) balance; however, comparisons across cohorts remain difficult because "imbalance" has been poorly defined. The Robotic Evaluation of Articular Laxity (REAL) classification was proposed to standardize ML imbalance using a 3 × 3 framework, but independent replication of this novel classification has not been performed. We retrospectively reviewed 386 robot-assisted TKAs for osteoarthritis. Medial and lateral extension and flexion gaps were recorded before bone resection and at procedure end. ML imbalance was calculated as signed and absolute values. Primary outcomes were the distribution and normality of signed ML imbalance and the goodness-of-fit of varus REAL category frequencies versus the original REAL report using thresholds of 2.5 and 5.0 mm. Exploratory analyses evaluated patient-reported outcomes across initial REAL categories. In the full cohort, initial signed ML imbalance deviated from normality, whereas varus initial extension imbalance was nearly normal (1.08 ± 2.79 mm). Final absolute ML imbalance was about 1 mm in extension and 1.0-1.3 mm in flexion, with significant paired reductions in both (both p < 0.001). Varus REAL category frequencies differed from the original report, but deviations were modest and concentrated in a subset of cells. Patient-reported outcomes did not differ across initial REAL categories. In this independent robot-assisted TKA cohort, key features of the REAL framework were replicated in varus knees. Intraoperative balancing substantially improved ML balance, supporting the descriptive utility of REAL for organizing intraoperative balance phenotypes across cohorts.