Rasch analysis of Dutch-translated version of the Foot Impact Scale for rheumatoid arthritis

Abstract

To translate the foot impact scale for RA (FIS-RA) to the dutch target language and to evaluate its internal construct validity using rasch analysis. Forward and backward translations of the original English version of the FIS-RA scale, combined with synthesis techniques and expert committee review, were undertaken to produce a final Dutch version with two subscales for impairment/footwear (FIS-RA(IF)) and activity/participation (FIS-RA(AP)). The pre-final version was field tested in RA patients to investigate face and content validity. FIS-RA questionnaires were completed by 207 Dutch RA patients. Rasch analysis tested the data for overall fit to the model, item and person fit, unidimensionality, differential item function (DIF) by age, gender and disease duration, targeting, reliability and local response dependency. Item deletion and re-analysis were planned, where Rasch model assumptions were violated. The FIS-RA(IF) (P < 0.0001) and FIS-RA(AP) (P < 0.0001) subscales did not fit the overall Rasch model. Misfitting items, DIF by age, gender and disease duration, and local response dependency were observed in both subscales. Item thresholds showed good coverage over both scales although a floor effect was observed for the FIS-RA(AP) subscale. The person separation index was 0.81 and 0.92 for the FIS-RA(IF) and FIS-RA(AP) subscales, respectively. Both subscales were not unidimensional. Item deletion and repeat Rasch analysis produced two subscales that fitted the Rasch model and were unidimensional. A Dutch language version of the FIS-RA questionnaire was successfully developed using Rasch analysis. Subscales for impairment/footwear and activity/participation showed good construct validity and were unidimensional.