Abstract
Background and purpose: Patient-reported outcome measures need to reflect true differences rather than measurement artifacts resulting from translations when scores are compared across countries. We evaluated the measurement invariance of the English, Dutch, Danish, and French Oxford Hip Score (OHS) and Oxford Knee Score (OKS) in patients undergoing hip and knee arthroplasty.Methods: OHS and OKS responses from patients undergoing primary hip or knee arthroplasty for osteoarthritis between 2019 and 2022 were included from national and single-center registries in the United Kingdom, the Netherlands, Denmark, and Switzerland. We evaluated unidimensionality, monotonicity (Hi > 0.3) and local independence (Yen Q3 > 0.20 above the average indicating independence) in each language version separately. Model fit was considered satisfactory when Comparative Fit Index (CFI) and Tucker–Lewis Index (TLI) were ≥ 0.95, root mean square error of approximation (RMSEA) ≤ 0.06, and standardized root mean square residual (SRMR) ≤ 0.08. Measurement invariance across languages was analyzed with multiple group—confirmatory factor analysis on full, random, and matched sample datasets.Results: Across the hip and knee cohorts, unidimensionality was acceptable in the English (n = 21,108 and 28,230), Dutch (n = 36,792 and 29,651), Danish (n = 815 and 1,015), and French (n = 590 and 459) versions. RMSEA ranged from 0.072 to 0.092 for OHS, and from 0.057 to 0.061 for OKS and other fit indices were acceptable. The multiple group-confirmatory factor analysis showed acceptable measurement invariance across languages for both scores, with changes in RMSEA < 0.15 and CFI < 0.1 across gradually more constrained models.Conclusion: Measurement invariance of the OHS and OKS was supported across the English, Dutch, Danish, and French versions, with multiple group-confirmatory factor analyses model fit within predefined acceptable limits, supporting cross-comparison of the different language versions of these instruments.