Assessing fatigue in late-midlife: increased scrutiny of the Multiple Fatigue Inventory-20 for community-dwelling subjects.

Abstract

Previous methods examining the Multiple Fatigue Inventory-20 (MFI-20) fatigue questionnaire have been limited to classical test theory, for example, factor analytic approaches. We employed modern test theory to further strengthen the construct validity of the MFI-20 fatigue in a sample of healthy late-midlife subjects. Five subdimensions of perceived fatigue were examined in n = 7,233 subjects: general fatigue, physical fatigue, reduced activity, reduced motivation, and mental fatigue. Fatigue burden was compared across age groups (aged 48-52 vs. 57-63) and gender. Mokken item response theory was used to investigate dimensionality, monotonicity, and invariant item ordering (IIO). In both age groups, as well as by gender, the Motivation domain presented with weak scalability, suggesting that caution be exercised when interpreting sum scores. For all groupings, the strongest scaling properties were observed in the General Fatigue domain. However, the General Fatigue domain did not meet the property of IIO. Two domains (for all groupings) did meet the minimum criteria for the property of IIO: Physical Fatigue and Activity. Introducing model parameters for items served to enhance the interpretive power of the MFI-20, allowing for the identification of the most optimal scales. Poorly performing items were more easily identified, and person ability was assessed more accurately.