Rasch analysis of the brief Michigan Hand Questionnaire in patients with thumb osteoarthritis

Abstract

Abstract Background The brief Michigan Hand Questionnaire (brief MHQ) is a 12-item self-reported measure of hand function for patients with hand disorders which has been validated using Classical Test Theory. Rasch analysis can provide more detailed psychometric information. The purpose of this Rasch analysis is to assess the psychometric properties of the brief MHQ for patients with thumb osteoarthritis, and to make recommendations for improvements to the questionnaire if needed.  Methods The Michigan Hand Questionnaire and demographic data were collected from 923 thumb osteoarthritis patients treated in specialized clinics for hand surgery and therapy in the Netherlands. Rasch analysis was performed on the 12 items of the brief MHQ using RUMM 2030 to assess the fit of the brief MHQ to the Rasch model. To determine fit, analysis of fit summary statistics, individual person fit and individual item fit were assessed. Threshold distributions were assessed to identify if any items required rescoring. The Person Separation Index was calculated to measure reliability of the questionnaire. Differential item functioning was assessed to identify item bias, and Principal Component Analysis was performed to identify unidimensionality and local dependence. Results The brief MHQ showed misfit (χ2 = 1312.5, p < 0.0001) with 6 items having disordered thresholds and 9 items requiring rescoring. After deleting 3 of the rescored items due to significant item fit residuals, the brief MHQ had an acceptable reliability (Cronbach’s alpha = 0.79). Misfit to the model (χ2 = 49.6, p = 0.0001), multidimensionality (10.2% of t-tests were significant), and item bias from non-uniform differential item functioning for 7 items across many person variables were still found. Conclusion Although no satisfactory solutions were found to correct the misfit to the Rasch model, it is recommended that the response options of the brief MHQ be rescored, and that items 6, 9 and 10 be removed. The lack of unidimensionality indicates that the items do not represent the singular construct of hand disability and that totalling the scores of the brief MHQ does not provide a valid measure of hand disability for people with thumb osteoarthritis. The 37-item Michigan Hand Questionnaire may provide a better assessment of hand disability for patients with thumb osteoarthritis.