Fatigue in Parkinson's disease: measurement properties of a generic and a condition-specific rating scale

High-quality fatigue rating scales are needed to advance the understanding of fatigue and determine the efficacy of interventions. Several fatigue scales are used in Parkinson's disease, but few have been tested using modern psychometric methodology (Rasch analysis).

To examine the measurement properties of the generic Functional Assessment of Chronic Illness Therapy-Fatigue (FACIT-F) scale and the condition-specific 16-item Parkinson Fatigue Scale (PFS-16) using Rasch analysis.

Postal survey data (n=150; 47% women; mean age 70 years) were Rasch analyzed. The PFS-16 scores were tested according to both the original polytomous and the suggested alternative dichotomized scoring methods.

The PFS-16 showed overall Rasch model fit, whereas the FACIT-F showed signs of misfit, which probably was the result of a sleepiness-related item and mixing of positively/negatively worded items. There was no differential item functioning by disease duration but by fatigue status (greater likelihood of needing to sleep or rest during the day among people classified as nonfatigued) in the PFS-16 and FACIT-F. However, this did not impact total score-based estimated person measures. Targeting and reliability (≥0.86) were good, but the dichotomized PFS-16 showed compromised measurement precision. Polytomous and dichotomized PFS-16 and FACIT-F scores identified six, three, and four statistically distinct sample strata, respectively.

We found general support for the measurement properties of both scales. However, polytomous PFS-16 scores exhibited advantages compared with dichotomous PFS-16 and FACIT-F scores. Dichotomization of item responses compromises measurement precision and the ability to separate people, and should be avoided.