Rasch analysis of the functional index for hand osteoarthritis (FIHOA) in patients with rhizarthrosis.

Abstract

To evaluate the measurement properties of the functional index for hand osteoarthritis (FIHOA) in subjects with rhizarthrosis using classical and modern psychometric methods. 164 individuals with rhizarthrosis were included in this retrospective study. We conducted a Confirmatory Factor Analysis (CFA) for unidimensionality, Cronbach's alpha (<i>α</i>) for internal consistency, and Rasch analysis (RA) for internal construct validity (stochastic invariance, monotonicity, local independence, unidimensionality, measurement invariance), reliability, and targeting. The FIHOA demonstrated sufficient internal consistency (<i>α</i> = 0.87). The final CFA partially fit to a unidimensional model (RMSEA = 0.08; SRMR = 0.06; CFI = 0.98; TLI = 0.97). Base RA revealed a fit to the model (<i>χ</i>²_df = 26.28₂₀; <i>p</i> = 0.16), satisfied the requirement of monotonicity (no disordered thresholds), unidimensionality (percentage of significant tests = 4.3%, lower bound confidence interval = 1.3%), but not the local independence and measurement invariance. After creating three clinically coherent subtests to account for the local dependence, the final Rasch analysis fit to the model (<i>χ</i>²_df = 11.19₂₄; <i>p</i> = 0.34), satisfying the requirements of the local independence, monotonicity, and measurement invariance (a non-significant DIF for age). Reliability indexes for the fit estimates were suitable for individual-level measurements (Distribution-Independent Person Separation Index = 0.96). After statistical modifications through Rasch analysis, the FIHOA turns out to be a unidimensional, valid, reliable, and invariant measure of activity limitation in patients with rhizarthrosis.