Variability of Visual Assessment in Determining Tibial Component Alignment in Total Ankle Arthroplasty.

Background: Accurate implant positioning is essential for long-term success in total ankle arthroplasty (TAA). Although patient-specific instrumentation (PSI) and advanced imaging have improved technical precision, intraoperative alignment assessment still relies heavily on visual interpretation of fluoroscopy. The reliability of visual assessment remains poorly understood. This study evaluated the accuracy and reliability of visual assessment of tibial component alignment in TAA. Methods: Seventeen raters assessed 21 TAA cases (15 unique, 6 duplicated) using standardized anteroposterior (AP) and lateral fluoroscopic or radiographic images. True implant alignment was defined by postoperative computed tomography (CT) measurements from a previously published PSI cohort. Raters classified angular alignment as neutral (0° to 2°), 2° to 4°, or greater than or equal to 5° of deviation in varus/valgus or dorsiflexion/plantarflexion. Interobserver agreement was assessed using Fleiss κ; and intraobserver reliability was evaluated using Cohen κ and intraclass correlation coefficients (ICCs). Accuracy, mean absolute error, and perceptual thresholds were analyzed using mixed-effects models and segmented logistic regression. Results: Overall classification accuracy was modest (48.3% AP; 52.5% lateral), with low interobserver agreement (κ = 0.14-0.21). Mean absolute error ranged from 2.0° to 2.2°. Intraobserver repeatability demonstrated moderate agreement (ICC(2,1) 0.43-0.48). Observer performance was most variable near neutral alignment and improved with larger deviations. Years of experience did not improve accuracy and demonstrated a small inverse association with correct classification. Conclusion: Visual assessment of tibial component alignment in TAA demonstrated substantial variability, particularly when evaluating small angular deviations. Imaging characteristics, including view orientation and modality, were associated with measurement error, whereas surgeon experience had minimal impact on overall accuracy. These findings highlight the limitations of visual assessment under standard fluoroscopic and radiographic imaging conditions and support continued investigation of other strategies, such as computer-assisted navigation, 3-dimensional imaging, and robotic-assisted technologies, to improve intraoperative alignment assessment. Level of Evidence: Level III, diagnostic reliability study.