ABSTRACT
OBJECTIVES: This study aimed to assess the psychometric properties of two widely used fatigue scales in a sample of patients with inflammatory conditions. METHODS: Rasch analysis was used to examine scale reliability, item bias, unidimensionality and overall fit to the Rasch model. Sub-test methodology was utilised to attempt to improve model fit for the CFQ and BRAF-MDQ. RESULTS: Initial analysis displayed strong reliability (PSI=0.89 0.96), alongside a lack of item bias in both scales. However, evidence for unidimensionality was not found for either scale. Overall fit to the Rasch model was marginal for the CFQ, and misfitting for the BRAF-MDQ. Local dependency was observed, as well as significant item misfit for both scales. Sub-test modifications resulted in the best model fit for the BRAF-MDQ (χ2(16)=15.77, p=0.469) and the CFQ (χ2(25)=15.49, p=0.929). Modifications resulted in improved fit, reductions in measurement error, and the production of ordinal-to-interval conversion tables for both scales. Conversion tables apply the benefits of enhanced measurement accuracy, valid comparison of BRAF-MDQ and CFQ scores to other interval-level data, appropriate use in parametric statistics, and enhanced precision in clinical cut-off scores-without the need to change administration format. CONCLUSION: The BRAF-MDQ and CFQ are valid, reliable tools for fatigue assessment. Psychometric indices and content factors suggest the CFQ is suited to measuring general fatigue, particularly when response burden is a concern, while the BRAF-MDQ should be used in clinical presentations where other symptoms are severe and the impact of fatigue on daily living, emotional, and social well-being is of interest.
